Intel Core i7 Processors [models & pricing] ! !

[GoriLLakoS]

Hello everybody :)

I searched a little in order to see if there are same posts arround but google did not saw me nothing except the Core i7-940.

So, the first releases in 2 months from the intel side will be the following :

1. Intel Core i7 Extreme Edition i7-965(3.2GHz,8MB L3, LGA1366, Product Code BX80601965 & QPI 6.4GT/sec) Estimated Price in EU : ~1000€

2. Intel Core i7-940(2.93GHz,8MB L3, LGA1366, Product Code BX80601940 & QPI 4.8GT/sec) Estimated Price in EU : ~510€

3. Intel Core i7-920 (2.66GHz,8MB L3, LGA1366, Product Code BX80601920 & QPI 4.8GT/sec) Estimated Price in EU : ~260€


Source (http://www.hwbox.gr/showthread.php?t=2300)

[xoqolatl]

Is that the regular price tiers of 1000$, 510$ and 266$ adjusted for EU prices, or is it more?

[Nosfer@tu]

Not as bad as it could be.

Normaly they are released at 1100-1250 If I remember correct.

[Jamesrt2004]

Nice :D good info :)

[largon]

I wonder if i7-920 could reach 3GHz. Reference clk from 133MHz to 150MHz might be pushing it...

[TedShred]

Interesting. One Euro = $1.45 or so US, so those prices are higher than earlier estimates of $999, $562 and $284. I'm not sure of the reliability of the site linked below (or its quoted prices), but I'm positive I'd seen it posted here before?

http://en.hardspell.com/doc/showcont.asp?news_id=3734

[RPGWiZaRD]

Is it only me that isn't very excited about this release? If overclocking without unlocked multi is so bad as some here have hinted, I see no reason to pick up one of these over a Wolfdale/Yorkfield anytime soon. Hopefully York gets cheaper though, that would be awesome news.

[Hornet331]

prices in europe are with vat included.

QX9770 is 1100€ with 19% vat and the QX9650 is 780€ with 19%vat, so i think we might see the same for the Ci7 965. A max of 1100-1200€.

[TedShred]

Is it only me that isn't very excited about this release? If overclocking without unlocked multi is so bad as some here have hinted, I see no reason to pick up one of these over a Wolfdale/Yorkfield anytime soon. Hopefully York gets cheaper though, that would be awesome news.

Yeah, at this point I'm waitin' and seein' and feeling ok about sticking with the current platform for awhile yet. It's cool to follow the story though, it'll be really interesting to see all this stuff live and in retail action.

edit: ah yes, VAT (19%?! holy smokes)

[bowman]

Is it only me that isn't very excited about this release? If overclocking without unlocked multi is so bad as some here have hinted, I see no reason to pick up one of these over a Wolfdale/Yorkfield anytime soon. Hopefully York gets cheaper though, that would be awesome news.

I was very psyched until info leaked that they will clock like a dead dog. I'm definitely taking a wait'n'see approach to see what the 2.66 version will do.

[STaRGaZeR]

i7-920 here I come. Cheaper than my old E6600 at launch :D :D

[Blauhung]

I wonder if i7-920 could reach 3GHz. Reference clk from 133MHz to 150MHz might be pushing it...

also remember that after tweaking with the Turbo mode settings you can pretty much add on a constant extra 2 points to the multi. So that 150 would end up pushing the CPU to 3.3GHz

[GoriLLakoS]

In the prices i have already included the 19% VAT.

I guess for the XE model that will be as the other XEs and they will start from 1000 to 1300 depending the country.

The other 2 prices will be there 2-3 weeks after the release :)

[freeloader]

also remember that after tweaking with the Turbo mode settings you can pretty much add on a constant extra 2 points to the multi. So that 150 would end up pushing the CPU to 3.3GHz

Correction....you'd be pushing 1 core to 3.3ghz, not all of them.

[TedShred]

actually, from the "on second thought" dept:

I've been playing around with the idea of tri-SLI but, from my (admittedly skinny) research it seems there's a lot of peeps not too happy with the NVidia chipsets. I understand the X58 supports SLI, so I'm really curious to see how that set up works. If it's solid, it would be tempting to go I7 if I end up going with NVidia/SLI.

[Donnie27]

In the prices i have already included the 19% VAT.

I guess for the XE model that will be as the other XEs and they will start from 1000 to 1300 depending the country.

The other 2 prices will be there 2-3 weeks after the release :)

Not really. If any manufacturer says Tray Price for a 1000 is $999, that's what it is. Then distributers and VAR's can sell them for whatever the heck they think you'll pay.

Not as bad as it could be.

Normaly they are released at 1100-1250 If I remember correct.

As seen by from Newegg.com when it was $1599 for a X6800 while most other sites still listed them for $1050 to $1100. TP was $999 while that was going on.

[Shintai]

Correction....you'd be pushing 1 core to 3.3ghz, not all of them.

No, you can turbo all 4 cores if the TDP headroom is there on your cooling solution.

I think you mix it up with the mobile 45nm Core 2 variant of it.

[bowman]

Correction....you'd be pushing 1 core to 3.3ghz, not all of them.

AFAIK the Turbo Mode can affect one, several or all cores depending on situation. I assume Blauhung refers to some mobo manufacturer's clever BIOS trick to force a constant Turbo Mode effect, or something.

Given that the E0 Penryn quads can reach 4ghz on air nowadays I'm less and less tempted by this.

[jmke]

Interesting. One Euro = $1.45 or so US, so those prices are higher than earlier estimates of $999, $562 and $284. I'm not sure of the reliability of the site linked below (or its quoted prices), but I'm positive I'd seen it posted here before?

http://en.hardspell.com/doc/showcont.asp?news_id=3734

actually those prices for US stand, EU just get screwed;)

[@@@@]

I wish retailer don't jack up the price so much and also it is kind of misleading $999, $562, $284 is for 1000 quantity we have to include TAX and of course mark up from wholesalers to retailers so basically it is different for every location around the world

[STaRGaZeR]

I assume Blauhung refers to some mobo manufacturer's clever BIOS trick to force a constant Turbo Mode effect, or something.

Is not constant Turbo Mode, is that you can use the 2 multipliers the mobo uses for Turbo like a normal multi, so for overclocking effectively the multi in these beast is up to its rated one + 2.

actually those prices for US stand, EU just get screwed;)

Like always. No surprise. Still dirt cheap for a CPU with that perfomance compared to current Penryn.

[Shintai]

actually those prices for US stand, EU just get screwed;)

Its retail prices and its with 19% VAT. And its on a new product. Where is the gounging!! Newegg would be in shock :rofl:

[shiznit93]

also remember that after tweaking with the Turbo mode settings you can pretty much add on a constant extra 2 points to the multi. So that 150 would end up pushing the CPU to 3.3GHz
so are you conceding that the reference clock is limited? If the 2.6 nehalem can't hit 3.6 or better I don't see a point in getting it over a current quad unless one can use all 8 threads effectively.

[jaredpace]

The 8mb cache is L3 right?

[m0da]

any reason the 920 and 940 have the same product code? or is the 920 just BX80601920?

[Blauhung]

so are you conceding that the reference clock is limited? If the 2.6 nehalem can't hit 3.6 or better I don't see a point in getting it over a current quad unless one can use all 8 threads effectively.

haha, like I have any clue how high the reference clock goes.

I just make the chips, they don't let me play with them:rofl:

But yeah, as Dr. Who was talking about in another thread. Turbo mode is dynamic overclocking based on how the chip sees itself using it's current thermal headroom. If you make the chip think it has much greater room, assuming you have cooling capable of dissipating the heat, then turbo mode will be active on all cores as long as they're being utilized.

[villa1n]

so are you conceding that the reference clock is limited? If the 2.6 nehalem can't hit 3.6 or better I don't see a point in getting it over a current quad unless one can use all 8 threads effectively.

So in a round about way they re saying 150 ref clock is a reasonable oc..so 3.0+200-300 mhz from turbo tweak...near constant, unless your loading all 4 cores? so lets avg it 3.2ghz, at 284$ usd.. puts it up against a q9450, which can hit 4.0 if you can hold a 500fsb, which doesnt seem as hard as it used to be.. Its gonna be pretty close, it ll come down to what kind of programs your using... mainly single threaded or multiples which will take advantage of smt.

[GoriLLakoS]

Sorry for the typos on the 1st post.

L3 is 8MB not L2 , i just correct it
The same with the product code of 920 nad 940.

It seems i am in love :p

[XS Janus]

So how long till I need to pay 260€ for something I dont really need? :D
2 months you say....?

[TedShred]

So how long till I need to pay 260€ for something I dont really need? :D
2 months you say....?

:rofl::rofl::rofl::rofl:

[bowman]

If you make the chip think it has much greater room, assuming you have cooling capable of dissipating the heat, then turbo mode will be active on all cores as long as they're being utilized.

Time for BIOS makers to implement the 'ultra OC mode' that makes the processor think it's in the middle of space. :rofl: Jokes aside it should be possible to fool that PCU-thingy for some tweaking fun..

[Silaz]

So, does anyone know when this is to be released? It would be nice to know the speculated month atleast.

[AuDioFreaK39]

So, does anyone know when this is to be released? It would be nice to know the speculated month atleast.

I've been tracking Nehalem for months. Here are the latest and most recurring speculative releases:


week 40 (September 29 - October 5, 2008)
week 44 (October 27 - November 2, 2008)
first two weeks of November


In my opinion, the first two weeks of November seems most likely. Afterall, the first Penryn (QX9650) was released on November 12, 2007.

[gOJDO]

The 920 looks very attractive for the price. The big question now is: can it run above 3.8GHz stable 24/7(OCed as much as my Q9450)?

Γορίλες, ευχαριστώ για την είδηση. :toast:

[AuDioFreaK39]

The 920 looks very attractive for the price. The big question now is: can it run above 3.8GHz stable 24/7(OCed as much as my Q9450)?

Γορίλες, ευχαριστώ για την είδηση. :toast:

Doubt it. I wouldn't expect it to get past 3.3GHz, but someone please tell me otherwise because that's still pretty unfortunate for 920 buyers.

[LaMpiR]

What would be difference from L3 and L2 cache?
I think I read an article saying that overclocking Nehalem is almost good as having second graphic card. Why wouldn't it able to oc that much?

[gOJDO]

What would be difference from L3 and L2 cache?compared to what?
I think I read an article saying that overclocking Nehalem is almost good as having second graphic card. You mean, you red a title?
Anyway it sounds like a BS to me.
Why wouldn't it able to oc that much?Much more complex logic, larger die, new m-arch, new stepping, higher power consumption and higher generated heat. It will OC better with the newer steppings and as the production process matures.

[LaMpiR]

I was thinking about difference between L2 vs L3?

[zanzabar]

L3 is much slower and higher latency than l2, and l3 is used a buffer for the memory controller were l2 is typicaly for cpu cash (things that need to be processed), L3 aslo is for the whole cpu and not per core like L1/L2

[LaMpiR]

Ok. TNx for clarifying that for me.

[villa1n]

The 920 looks very attractive for the price. The big question now is: can it run above 3.8GHz stable 24/7(OCed as much as my Q9450)?

Γορίλες, ευχαριστώ για την είδηση. :toast:

If it can oc to 3.3 -3.4 + turbo, it should match or beat a 9450....at least thats my hope, otherwise, 9550 it for me :P

[T_Flight]

One thing alot of people are fortgetting is the technology behind this CPU. You can't compare this CPU to the old FSB models. The FSB is what limited performance in many ways. It's been the bottleneck for memory for ages. It is woefully ineficient comapred to the QPI this chip has.

All cores are now linked via the QPI. The mem controller is now on the CPU which gets rid of latency issues. The boardas for these new CPU's will be leaps and bounds beyond what the older boards were. A P6T for example has 16 phase power! An 8 layer board with 16 phase power from Asus. There is nothing like that that's remotely in the realm of that board.

We're going to Tri Channel, and a new socket. This is a complete upgrade and will give a path for future upgrades where the 775 based systems are gonna become obsolete and will have less and less of an upgrade path as time goes on.

I am most definitely going with this technology. Not just becasue it's new, but becasue it will have at least some measure of having the ability to do future upgrades. It will excel in the deamanding applications of video encoding and editing, my flight simulators, and folding. Will it make much a difference in Internet surfing? I doubt we'd notice it, but in the deamanding apps, and in the new software that will be developed this CPU will shine.

We're in the Vista era now, and Win7 is being talked about. The Conroe systems were a new breed, but they are switching gears once again, and this CPU is gonna excel. It's an Intel...what else would one expect? :)

[Donnie27]

haha, like I have any clue how high the reference clock goes.

I just make the chips, they don't let me play with them:rofl:

But yeah, as Dr. Who was talking about in another thread. Turbo mode is dynamic overclocking based on how the chip sees itself using it's current thermal headroom. If you make the chip think it has much greater room, assuming you have cooling capable of dissipating the heat, then turbo mode will be active on all cores as long as they're being utilized.

Sweet now that's what I wanted to hear.

[Donnie27]

One thing alot of people are fortgetting is the technology behind this CPU. You can't compare this CPU to the old FSB models. The FSB is what limited performance in many ways. It's been the bottleneck for memory for ages. It is woefully ineficient comapred to the QPI this chip has.

FSB limitations don't show themselves until you move past Dual Sockets. On the Desktop, A64 and X2 did the nasty to Intel because it had a better architecture for its core, not an IMC.

All cores are now linked via the QPI. The mem controller is now on the CPU which gets rid of latency issues. The boardas for these new CPU's will be leaps and bounds beyond what the older boards were. A P6T for example has 16 phase power! An 8 layer board with 16 phase power from Asus. There is nothing like that that's remotely in the realm of that board.

If it needs 16 phase, it might NOT be the most effiecent thing out there. I want to use less, not more. I know the IMC and etc... has to be figured in.

We're going to Tri Channel, and a new socket. This is a complete upgrade and will give a path for future upgrades where the 775 based systems are gonna become obsolete and will have less and less of an upgrade path as time goes on.

Memory isn't really stressed out right now and Memory Bandwidth goes wasted too many times. It hasn't helped AMD since Conroe shipped? Tri-Channel right now might be more ***like RAMBUS was on the P3. I hope I'm wrong. I know enough from at lest 4 different folks that overclocking will not be a problem for most folks here. Again, if none serious overclockers like Tom's Hardware can get 4GHz, worries about it not overclocking is worrying about nothing.

I am most definitely going with this technology. Not just becasue it's new, but becasue it will have at least some measure of having the ability to do future upgrades. It will excel in the deamanding applications of video encoding and editing, my flight simulators, and folding. Will it make much a difference in Internet surfing? I doubt we'd notice it, but in the deamanding apps, and in the new software that will be developed this CPU will shine.

We're in the Vista era now, and Win7 is being talked about. The Conroe systems were a new breed, but they are switching gears once again, and this CPU is gonna excel. It's an Intel...what else would one expect? :)

Good Luck!

[ragzarok]

One thing alot of people are fortgetting is the technology behind this CPU. You can't compare this CPU to the old FSB models. The FSB is what limited performance in many ways. It's been the bottleneck for memory for ages. It is woefully ineficient comapred to the QPI this chip has.

All cores are now linked via the QPI. The mem controller is now on the CPU which gets rid of latency issues. The boardas for these new CPU's will be leaps and bounds beyond what the older boards were. A P6T for example has 16 phase power! An 8 layer board with 16 phase power from Asus. There is nothing like that that's remotely in the realm of that board.

We're going to Tri Channel, and a new socket. This is a complete upgrade and will give a path for future upgrades where the 775 based systems are gonna become obsolete and will have less and less of an upgrade path as time goes on.

I am most definitely going with this technology. Not just becasue it's new, but becasue it will have at least some measure of having the ability to do future upgrades. It will excel in the deamanding applications of video encoding and editing, my flight simulators, and folding. Will it make much a difference in Internet surfing? I doubt we'd notice it, but in the deamanding apps, and in the new software that will be developed this CPU will shine.

We're in the Vista era now, and Win7 is being talked about. The Conroe systems were a new breed, but they are switching gears once again, and this CPU is gonna excel. It's an Intel...what else would one expect? :)

Nope not buying it.
I can't tell you how many times I've upgraded board, card and CPU, and RAM, in the reign of 775 Socket. As always, incremental changes and optimizations are bound to come as the architecture matures. It is my opinion that now is the time to hop on that high end or performance 775 train and ride it out for about a year or year and a half till the price comes down, the technology matures, and the software catches up to it.

[AnXioZ]

haha, like I have any clue how high the reference clock goes.

I just make the chips, they don't let me play with them:rofl:

But yeah, as Dr. Who was talking about in another thread. Turbo mode is dynamic overclocking based on how the chip sees itself using it's current thermal headroom. If you make the chip think it has much greater room, assuming you have cooling capable of dissipating the heat, then turbo mode will be active on all cores as long as they're being utilized.
If that is the case, then there will be a lot of misleading reviews circling around the web. I guess they'll use stock cooling, but still results will vary as temperatures vary. Talk about confusion :confused:
Nope not buying it.
I can't tell you how many times I've upgraded board, card and CPU, and RAM, in the reign of 775 Socket. As always, incremental changes and optimizations are bound to come as the architecture matures. It is my opinion that now is the time to hop on that high end or performance 775 train and ride it out for about a year or year and a half till the price comes down, the technology matures, and the software catches up to it.
That's a very strong point, but if you can get better price/performance ratio on i7 why wouldn't you let the technology wave drown you?

[Klarko]

Question One: Ive heard these OC really well, any truth to that?

Question Two: I've heard i7 doesn't really make gaming performance increase, and if that's true would it be better to just stick with the Q9550 if that was my other option? [pending i would pick the 2.66 or 2.93]

[Hornet331]

I dont really care about the ocing of nehalem right now, but i doubt it will be much worse then C2D. Since im plaining to go for the XE version anyway.

But the biggest thing im looking forward to is 8 threads on one cpu, hell yeah boinc will pay off with that. :D

[dinos22]

sheesh 1K+ Euros hey heh eek




haha, like I have any clue how high the reference clock goes.

I just make the chips, they don't let me play with them:rofl:

But yeah, as Dr. Who was talking about in another thread. Turbo mode is dynamic overclocking based on how the chip sees itself using it's current thermal headroom. If you make the chip think it has much greater room, assuming you have cooling capable of dissipating the heat, then turbo mode will be active on all cores as long as they're being utilized.

well that's just not cool lol

[Nasgul]

The same with the product code of 920 nad 940.
oh-oh!

Flash back!

Pentium D 920 and Pentium D 940. :ROTF:

I luv Preslers.

Anyhow, the price for the 2.66ghz doesn't look bad at all.....

4 core and HTT? Seck-xy!!!!

[saint-francis]

Nope not buying it.
I can't tell you how many times I've upgraded board, card and CPU, and RAM, in the reign of 775 Socket. As always, incremental changes and optimizations are bound to come as the architecture matures. It is my opinion that now is the time to hop on that high end or performance 775 train and ride it out for about a year or year and a half till the price comes down, the technology matures, and the software catches up to it.

I agree. You know that the motherboards will change drastically in the first year if nothing else even of the processors don't (highly doubtful). Remember the cheesy boards we had to deal with when Conroe hit the market? That was only 2 years ago. I figure I'll continue to run my Q6600 into the ground and when it's finally hosed I'll get some 45 nm quad for short money. When that's toast or I think that the Nehalem market is stabilizing I'm going to make a powerhouse with one of them. Then I'm going to treat it as a server; offload all of my video encoding and other busy work onto it. I'll keep my current board as a work station and fill it up wit a RAID array of SSD's. Who needs a work station with any more power than a Q6600 has? For the day to day tasks what you need is a decent amount of processing power, a good amount of good RAM and the about 10 SSD's in RAID 5. ;)

[Donnie27]

I agree. You know that the motherboards will change drastically in the first year if nothing else even of the processors don't (highly doubtful). Remember the cheesy boards we had to deal with when Conroe hit the market? That was only 2 years ago. I figure I'll continue to run my Q6600 into the ground and when it's finally hosed I'll get some 45 nm quad for short money. When that's toast or I think that the Nehalem market is stabilizing I'm going to make a powerhouse with one of them. Then I'm going to treat it as a server; offload all of my video encoding and other busy work onto it. I'll keep my current board as a work station and fill it up wit a RAID array of SSD's. Who needs a work station with any more power than a Q6600 has? For the day to day tasks what you need is a decent amount of processing power, a good amount of good RAM and the about 10 SSD's in RAID 5. ;)

Cheesy? There wasn't anything wrong with i965 and Conroe ran on last revision of i975 as well. Conroe was a new generation but Nehalem will not suffer from the slow but enough support Conroe got. Too much money to be made.

I'm really sick of folks talking about $1000 CPU's when 2.66GHz might cost about $300 or so. Or less than Conroe E6600 or Athlon X2 3800+. Get real folks what you guys want?Now yield must be danged good if 2.66GHz is the slowest they're going to ship:up: Or has some one mentioned this already?

[xVeinx]

Having had a (intel dp965lt) P965 board and E6300 a month or so after release, I can honestly say that the drivers for that board were less than stellar. It took months for them to get things working smoothly. Basic things like getting the computer to shut down properly didn't always work. I'm not entirely complaining, as I enjoyed the problem solving and such, but realistically the board needed work. I'm not going to naysay Intel's ability to get it right, but I think it reasonable to say that the convergence of a new memory design, new system bus, new socket, new chipset, and new core all at once is a large bit. It's going to take time for manufacturers to get the drivers, bioses, etc. correct. My prediction is that we'll see marginal performance increases in memory speed, etc from the system as well as I/O issues until the quirks get worked out. Is the potential for the architecture impressive? Sure. Just don't expect it to be smooth going for the first few months. They don't call it the bleeding edge for nothing...

EDIT: This was seen even more recently with the Intel X38/X48 skulltrail board. Not to pick on Intel boards, as others have had issues as well, but there are still quirks that haven't fully been solved.

[saint-francis]

Don't get me wrong! I got a DFI 975X/G and an E6600 right off the bat and I was in love. I had a %50 OC Orthos stable in about 1/2 hour. But my DFI P35 and my Q6600 beat the bag out of my first C2D system. Obviously the new lineup is going to be stellar. I'm just personally going to wait until everything gets comfortable before I jump on the i7 train.

[RPGWiZaRD]

Nope not buying it.
I can't tell you how many times I've upgraded board, card and CPU, and RAM, in the reign of 775 Socket. As always, incremental changes and optimizations are bound to come as the architecture matures. It is my opinion that now is the time to hop on that high end or performance 775 train and ride it out for about a year or year and a half till the price comes down, the technology matures, and the software catches up to it.

That's my opinion as well, I'd rather stay with a good socket 775 setup for a while until this new technology matures and gets cheaper. Also DDR3 pricing and offerings ain't on the level yet I'd like it to be. I'm sure in 1 year it'll better.

[paulhamm]

Nope not buying it.
I can't tell you how many times I've upgraded board, card and CPU, and RAM, in the reign of 775 Socket. As always, incremental changes and optimizations are bound to come as the architecture matures. It is my opinion that now is the time to hop on that high end or performance 775 train and ride it out for about a year or year and a half till the price comes down, the technology matures, and the software catches up to it.

That's my opinion as well, I'd rather stay with a good socket 775 setup for a while until this new technology matures and gets cheaper. Also DDR3 pricing and offerings ain't on the level yet I'd like it to be. I'm sure in 1 year it'll better.

Still wondering about this also. Though I expect the 775 has peaked or close to it. I would like to get a deal on a nice quad for one of my DT systems. I figure when the core i7 hits the bulk of the 775 procs are going to start hitting fire sale prices.

[bowman]

Again, if none serious overclockers like Tom's Hardware can get 4GHz, worries about it not overclocking is worrying about nothing.

Yeah, but everything seems to indicate this is on the 'XE'.. What worries people (me along with them) is the OC-age on the non-silly ones.

[Epsilon84]

Still wondering about this also. Though I expect the 775 has peaked or close to it. I would like to get a deal on a nice quad for one of my DT systems. I figure when the core i7 hits the bulk of the 775 procs are going to start hitting fire sale prices.

Thats my plan anyway. I'm currently running a C2D, I'm waiting for the Q9650 price to drop after Nehalem launches. Gonna overclock it to around 4GHz (hopefully), should last me a good while.

[Kingcarcas]

Can't wait to see some world records with these chips.
of video encoding and editing, my flight simulators, and folding.
I like your tone :up:

[Booj]

I want to know if they coldbug or not. I saw that article @ Nordichardware saying one was running at -120c which is promising. If it turns out they like the cold.. then it's time to buy a cascade :D :D

[@@@@]

Same here plan on waiting for the price drop then decide which one to buy depending on the price drop, Nehalem is a great processor but i usually wait for the shrink derivative of the Architecture from their tick tock strategy, Westmere (formerly Nehalem-C) 32nm

[dinos22]

I want to know if they coldbug or not. I saw that article @ Nordichardware saying one was running at -120c which is promising. If it turns out they like the cold.. then it's time to buy a cascade :D :D

put your order in mate :D
you getting a kayl cascade? :d

[Booj]

put your order in mate :D
you getting a kayl cascade? :d

ehehe that's what I like to hear :D :up:

Yeah going for a Kayl unit. Not just yet though, have to spend a little bit on i7 EE, triple channel etc :rofl:

[Bo_Fox]

Guys, with the IMC, triple channel is not going to bring any benefits over dual channel, IMHO... since IMC should already get rid of the latency issues and memory bandwidth issues. Having to buy 3x DDR3 memory modules is gonna be very expensive for now anyways, and rather impractical. Who needs 6GB nowadays instead of 4GB which is already plenty?

Plus, who needs 8 virtual cores? Please give me a *popular* application that will really benefit from 8 "virtual" cores rather than 4 physical cores? I do not know of a game that would use 8 cores at all.. not even Supreme Commander might benefit from 8 "virtual" cores. Hyperthreading on virtual cores only gives like a maximum of 10% improvement over using just the physical cores after all, right? If I got a Nehalem, I'd just disable HT and use the 4 physical cores instead, since it would actually improve performance on most applications that do not care about using 8 virtual cores rather than 4 physical cores. It would probably be 3 years from now when any applications will actually benefit greatly from 8 virtual cores. Otherwise, why aren't you guys sticking to the 2-year-old Pentium 4 chips with Hyperthreading today when there are finally "numerous" games/apps to take advantage of it? 8 virtual cores.. BLAH!

[XS Janus]

Are we talking about games again...
What's wrong? Be happy you DON'T need to upgrade...

[Nedjo]

Bo Fox, you're absolutely right on both issues - tri-channel mem. ctrl. will only show it's strengths in SiSoft Sandra, and efficiency of HT is academic, according to Anand:

http://images.anandtech.com/reviews/cpu/intel/nehalem/uarch/HTperformance.jpg

I mean 10% in ideal paralleled situation like rendering is laughable!

[JPQY]

Well..i'am already waiting for many years when they come with cpu's with multi cores..i'am not a gamer but i play chess,do analyses and so on..and have never enough cores! There are chess engines ready for 2048cores,so i think i have to wait many years when i see a computer who has these cores!
But you are right,this is just one example off software and chess use these cores 100%.

Here some test results:

Engine ------------- Depth--- Time- Total nodes - Nodes/s.--Score ----- Hardware
------------------------------------------------------------------------------------------------------------------------------
1)TogaII142JD-8cpu 21/46 03:46 208.317.723 7.345.841 +0,26 Skulltrail 2xQX9775 @4.2Ghz DDR2: 4-5-5-17
2)TogaII141SE-16cpu 21/45 03:30 94.511.341 6.843.413 +0.22 Gainestown 2xNehalem @3.07Ghz DDR3: 9-9-9-24
3)TogaII142JD-16cpu 21/46 03:59 96.988.402 6.521.362 +0.26 Gainestown 2xNehalem @2.93Ghz DDR3: 8-8-8-19
4)TogaII142JD-8cpu 21/52 03:31 174.872.390 6.286.188 +0.25 Skulltrail 2xQX9775 @ 4.0Ghz
5)TogaII142JD-8cpu 21/46 03:54 172.582.215 5.891.624 +0.30 Nehalem @ 2.93Ghz
6)TogaII142JD-8cpu 21/50 03:37 154.531.874 5.278.434 +0.26 Skulltrail Harpertown 2xE5430 @ 3.0Ghz
7)TogaII141SE-4cpu 21/52 05:12 364.182.100 4.576.252 +0,20 Q6600 g0 @ 3.2Ghz (My PC) DDR2: 5-5-5-15
8)TogaII142JD-4cpu 21/52 03:48 260.594.079 4.556.842 +0,23 Q6600 g0 @ 3.2Ghz (My PC) DDR2: 5-5-5-15
9)TogaII142JD-8cpu 21/48 04:19 106.946.778 3.351.022 +0,20 2xOpteron 2347 @1.9Ghz

JP.

[Epsilon84]

Bo Fox, you're absolutely right on both issues - tri-channel mem. ctrl. will only show it's strengths in SiSoft Sandra, and efficiency of HT is academic, according to Anand:

http://images.anandtech.com/reviews/cpu/intel/nehalem/uarch/HTperformance.jpg

I mean 10% in ideal paralleled situation like rendering is laughable!

So you'd say no to 10 - 30% free performance? :rolleyes:

Even without HT Nehalem is already heads and shoulders above any existing CPU in multi-threaded performance, HT is just the icing on the cake.

I bet if AMD somehow incorporated HT and even got a modest 5 - 10% boost you wouldn't find it so 'laughable' or 'academic'. Just a hunch. ;)

[Nosfer@tu]

Nope not buying it.
I can't tell you how many times I've upgraded board, card and CPU, and RAM, in the reign of 775 Socket. As always, incremental changes and optimizations are bound to come as the architecture matures. It is my opinion that now is the time to hop on that high end or performance 775 train and ride it out for about a year or year and a half till the price comes down, the technology matures, and the software catches up to it.

new technoligy isent for people that want the best system for the buck!!!!!!

Cheesy? There wasn't anything wrong with i965 and Conroe ran on last revision of i975 as well. Conroe was a new generation but Nehalem will not suffer from the slow but enough support Conroe got. Too much money to be made.

I'm really sick of folks talking about $1000 CPU's when 2.66GHz might cost about $300 or so. Or less than Conroe E6600 or Athlon X2 3800+. Get real folks what you guys want?Now yield must be danged good if 2.66GHz is the slowest they're going to ship:up: Or has some one mentioned this already?

why are you sick of that Donnie ?
Who cares if the cpu that is listed at 1000 $ only runs 30% faster than the cpu listed at 500 $

THIS IS XTREMESYSTEMS. NOT BUGET SYSTEMS, Not best bang for buck...

I dont understand why everybody gets so agitated.
RELAX, Dont buy if you dont like the price ratio.
BE HAPPY somthing new is comming out so you can save even more on the Q9450. JESUS :D RELAX EVERYBODY :D

[jmke]

THIS IS XTREMESYSTEMS. NOT BUGET SYSTEMS, Not best bang for buck...


last time I checked, overclocking was/is still a major part of getting more performance out of your components, if they are expensive or not.
it all started out exactly by most bang for the buck, it's not because the market tells you overclocking is cool and costs much, you have to follow...

[Hornet331]

snip

lol one bloomfield @ 2,93 ghz is in the same range as a duaksocket QX9770 @ 4ghz. :eek:

now thats performance. :up:

[Bo_Fox]

So you'd say no to 10 - 30% free performance? :rolleyes:

Even without HT Nehalem is already heads and shoulders above any existing CPU in multi-threaded performance, HT is just the icing on the cake.

I bet if AMD somehow incorporated HT and even got a modest 5 - 10% boost you wouldn't find it so 'laughable' or 'academic'. Just a hunch. ;)

Well, your hunch got me wrong... I do not have a benchmarking fetish. If I were all about 3dsMax, I'd be thrilled to have the 10% performance increase thanks to 8 virtual cores.

Well, the reality is that HT actually *decreases* performance on most apps that do not take advantage of multiple cores. That is why it was recommended to turn off HT on many P4 CPU's back then in the hey days. Heck, I'd love to turn it off until there are enough apps taking advantage of 8 cores. Most newer apps only use up to 2 cores, let alone 4 cores. Plus HT is known to bring some compatibility issues with a number of software programs/drivers. Man, I do not want even a bit of a headache from coping with any of those potential issues. If Core i7 does not allow for disabling of HT in the BIOS, I'd be further discouraged from buying it.

[.Logic]

THIS IS XTREMESYSTEMS...

Ah yes... the age old 'THIS IS XTREMESYSTEMS...' mantra. How ignorant of me, I must remember to abandon all intelligence and logic in accordance with the forum name ;)

Seriously though, Bo Fox I can understand where you're coming from however, in order for computer tech to progress forward something has to come first (what good would a multi threaded application be without a multi core processor to run it well). At least you can turn HT off if you are so inclined...

[Epsilon84]

Well, your hunch got me wrong... I do not have a benchmarking fetish. If I were all about 3dsMax, I'd be thrilled to have the 10% performance increase thanks to 8 virtual cores.

My post wasn't directed at you... ;)

Well, the reality is that HT actually *decreases* performance on most apps that do not take advantage of multiple cores.

I don't think so...
http://www.anandtech.com/showdoc.aspx?i=1746&p=6
It looks like Intel was able to deliver on their claims, most users should have no problem leaving Hyper-Threading enabled as it won't reduce performance.
http://images.anandtech.com/reviews/cpu/intel/pentium4/3.06G/image002.gif

This chart pretty much invalidates the remainder of your post regarding HT...

[Macadamia]

So you'd say no to 10 - 30% free performance? :rolleyes:

Even without HT Nehalem is already heads and shoulders above any existing CPU in multi-threaded performance, HT is just the icing on the cake.

I bet if AMD somehow incorporated HT and even got a modest 5 - 10% boost you wouldn't find it so 'laughable' or 'academic'. Just a hunch. ;)

You're completely off point.


Nehalem's ESSENCE has been SMT. I was actually really stoked about it- there's apps that aren't too optimized like Maxwell or Vue that could need the extra power that I'd appreciate. This is laughable for something hyped to hell and back. Where's the 50%+ that was speculated due to "SMT using the wider architecture better"? It hardly even matters at this point!

It's now truth: Dunnington is going to roll over, squish and ra*e Bloomfield, a "markedly superior" architecture in the apps that matter.

[bowman]

You can't compare a quad-socket platform to a unisocket platform.. Maybe if that big, fat six-core die was headed to the 775 socket in an LGA package, but it's not.

[Macadamia]

Dunnington is a sign of things to come. Sorta like K8 vs Pentium-M, this time roles reversed. Ahem.

[bowman]

No, Dunnington is the last breath of an architecture about to be replaced. I'm not sure what your point is at all. :confused:

[Jacky]

No, Dunnington is the last breath of an architecture about to be replaced. I'm not sure what your point is at all. :confused:
Especially considering the die size of Dunnington is probably in the realm of 500mm² and it will be destroyed by Beckton (the only arch it should be compared to). His point is nill.

The SMT gains are everything but laughable. A little bit more appreciation guys, one single feature gives a bigger performance boost than one whole die shrink from intel normally would. What's wrong with it?

[Epsilon84]

It's now truth: Dunnington is going to roll over, squish and ra*e Bloomfield, a "markedly superior" architecture in the apps that matter.

Random?! :rolleyes:

What does Dunnington have to do with Core i7? One is a unisocket desktop chip that costs $300 - $1000, another is a multi-socket server chip that costs $1000 - $3000. I don't see the relevance.

[Nosfer@tu]

last time I checked, overclocking was/is still a major part of getting more performance out of your components, if they are expensive or not.
it all started out exactly by most bang for the buck, it's not because the market tells you overclocking is cool and costs much, you have to follow...

AGREE. But this thread is not turning at that. :)

This is the same old storry, EVERYBODY loves the new stuff. But instead of debating how it works, How it overclocks and so on the "WOW it is accualy more expensive that a Q9450" arrises.

I agree. I normaly dont have high end GFX cards because they are so expensive!

Im not sure if we are misunderstanding each other.
I just felt sick of the "Wow it dosent give med 30% more power for the same amount of money" argument :D

Ah yes... the age old 'THIS IS XTREMESYSTEMS...' mantra. How ignorant of me, I must remember to abandon all intelligence and logic in accordance with the forum name ;)

Seriously though, Bo Fox I can understand where you're coming from however, in order for computer tech to progress forward something has to come first (what good would a multi threaded application be without a multi core processor to run it well). At least you can turn HT off if you are so inclined...

well I said what I mean above :)

I understand what everybody is saying. And I dont have 1000 Euro to spend my self.

I just think the thread was turning the wrong way :)
not that I have to decide, But I wanted to influence it in another direction :)

[Donnie27]

new technoligy isent for people that want the best system for the buck!!!!!!



why are you sick of that Donnie ?
Who cares if the cpu that is listed at 1000 $ only runs 30% faster than the cpu listed at 500 $

THIS IS XTREMESYSTEMS. NOT BUGET SYSTEMS, Not best bang for buck...

I dont understand why everybody gets so agitated.
RELAX, Dont buy if you dont like the price ratio.
BE HAPPY somthing new is comming out so you can save even more on the Q9450. JESUS :D RELAX EVERYBODY :D

Because folks will do just like they did with E6600 and NOT X6800 that got out numbered 10 to 1. A lot fewer folks even care about the $1000 dollar model. This is Xtreme Systems.org and many know higher overclocking room with lower end processors than those who buy the top models and do the same.

I was just saying that prices are up to the Distributors, OEMs and VAR, not just AMD or Intel. The price you mentioned wasn't an Intel price as you said. I didn't mean anything nasty or negative.

why are you sick of that Donnie ?
Who cares if the cpu that is listed at 1000 $ only runs 30% faster than the cpu listed at 500 $

THIS IS XTREMESYSTEMS. NOT BUGET SYSTEMS, Not best bang for buck...

I dont understand why everybody gets so agitated.
RELAX, Dont buy if you dont like the price ratio.
BE HAPPY somthing new is comming out so you can save even more on the Q9450. JESUS RELAX EVERYBODY

This isn't Xtremely Rich Folks.org either LOL! E6600 and Q6600 and now 9300, 9450 and 8400 seems to be the most used here. One of the aims of overclocking is adding Value by overclocking. Pay for a lowend model and end up with a higher end model.

Or am I wrong? But that wasn't the point. I still see folks giving Newegg's price and not Intel's Tray Price as I said.

[jaredpace]

lol one bloomfield @ 2,93 ghz is in the same range as a duaksocket QX9770 @ 4ghz. :eek:

now thats performance. :up:

i bet a 4.2ghz wolfdale is faster than a 2.93ghz nehalem in games:p:

[STaRGaZeR]

i bet a 4.2ghz wolfdale is faster than a 2.93ghz nehalem in games:p:

And a 4-4.2GHz Nehalem? ;)

[RaZz!]

i'd say quadcore penryns will do just fine in future games. atm, games aren't fully utilizing todays cpus at all - they're far more gpu- than cpu-dependant.

where nehalem really let's his muscle speak is in heavily paralleled apps like rendering, encoding and such.

[mikeyakame]

I am looking forward to being able to throw x264 encoder on the 8 virtual cores since it scales nearly 100% up to 16 cores at present. You can never encode 1920x1080 with all the good stuff turned up to their limits too fast ;> Hell if the 3.2GHz EE knocks off even 1 or 2 hours off a 24+ hour encode it's definitely worth every penny its selling for and I for one will be praising it. Any gain beyond the current limits of the C2D SP m-arch without going to DP server setups at present is welcomed by me.

All you guys complaining about performance in games and single threaded apps need to lighten up. This is an m-arch aimed at the enterprise / scientific / mathematics / video etc markets, IMC and HT has very little gains on regular home setups and probably won't be worth while for a lot of people until prices drop. For those which it isn't enjoy the price drops on the current C2D 45Nm line-up and game/whatever else you do on ;>

[shiznit93]

i'd say quadcore penryns will do just fine in future games. atm, games aren't fully utilizing todays cpus at all - they're far more gpu- than cpu-dependant.

where nehalem really let's his muscle speak is in heavily paralleled apps like rendering, encoding and such.
try warhammer online, I've never seen a more cpu dependent game including source engine games. even with my E8400@4.2ghz, the framerate at 1280x800 and 1920x1200 is identical.

[bowman]

And a 4-4.2GHz Nehalem? ;)

Good luck with that, unless you're on the XE. I don't care what the five people on here buying XEs think, I'm not spending $999 on a CPU to get a decent overclock when I can get a multiplier-locked Penryn instead which will fall in price when these are released..

If the 2.66GHz version is able to reach 4GHz or close to it, then sure. Otherwise, there's no point, Penryn is better.

[STaRGaZeR]

Good luck with that, unless you're on the XE. I don't care what the five people on here buying XEs think, I'm not spending $999 on a CPU to get a decent overclock when I can get a multiplier-locked Penryn instead which will fall in price when these are released..

If the 2.66GHz version is able to reach 4GHz or close to it, then sure. Otherwise, there's no point, Penryn is better.

I'm not going to argue with you because neither you or me know anything about overclocking headroom and clock per clock perfomance of Nehalem vs Penryn in various games. You can't extract any useful OC or perfomance data from the various previews we've seen with non-retail, no fully working mobo and early stepping Nehalems. Wait till they hit the streets for your judgement.

Me buying a XE? No way, I usually don't like to throw away my money ;)

[bowman]

You can't extract any useful OC or perfomance data from the various previews we've seen with non-retail, no fully working mobo and early stepping Nehalems. Wait till they hit the streets for your judgement.

Hey, nothing is final obviously, they haven't even officially announced these yet. I'll give them the benefit of the doubt, but things truly look sour currently.

[jaredpace]

yes the new cpu will be great for things like:

the enterprise / scientific / mathematics / video etc markets, IMC and HT

and only so-so for stuffs like:

gains on regular home setups and
performance in games and single threaded apps

So, I probably won't be getting one:(

[T_Flight]

try warhammer online, I've never seen a more cpu dependent game including source engine games. even with my E8400@4.2ghz, the framerate at 1280x800 and 1920x1200 is identical.


I know a few. Really they are not games. They fall into the sim category.

Falcon 4, Fighter Ops when it's released and many other sims are highly CPU dependent.

[STaRGaZeR]

Hey, nothing is final obviously, they haven't even officially announced these yet. I'll give them the benefit of the doubt, but things truly look sour currently.

Add to that what Drwho? has said here that there is an abismal difference between B0 stepping (the reviewers have this one) and the retail stepping. The only thing I don't like so far is the unconfirmed RAM-core voltage linking, that looks fugly. But again benefit of the doubt till the processor hits the streets.

[Metroid]

Drastic changes must be taken seriously, if not we would still be running slow processors forever. So what would be the point of getting a new improved working done?

It seems companies are holding back in this area for its factual conspiracy and undeniable casualties in the market. It is amazing we can get a processor who has 4 cores and 4 virtual working threads for a price like $300 that will be deployed at 2.6 Ghz. It is a breakthrough even if it can not be an overclocker dream. It is already proved by some known and credible technology websites that a 25% of improvement clock per clock is somewhat to be classified as a matured architecture. So I do not see the point of non getting it as soon as it is available for the majority of IT groups in the works.

Metroid.

[LaMpiR]

All of these are plain speculations... We can discuss it properly only when we see some numbers :)

[Bo_Fox]

My post wasn't directed at you... ;)



I don't think so...
http://www.anandtech.com/showdoc.aspx?i=1746&p=6

http://images.anandtech.com/reviews/cpu/intel/pentium4/3.06G/image002.gif

This chart pretty much invalidates the remainder of your post regarding HT...

This chart pretty much validates the fact that enabling HT *decreases* performance, if you look at CCWS2K2. I remember there being quite a number of other applications where performance is significantly decreased with HT. Anand only shows one but there was a site that showed a lot more and actually cautioned against HT if those applications were to be used (and even showed how it BSOD'ed in some applications). Heck, do you really use 3dsMax or Divx encoding? If not, then it's hardly worth it, IMHO.

[foch3 -USA-]

Wow! look at those Virtual cores just crushing.

[[XC] gomeler]

I'm certain Nehalem will offer higher IPC and efficiency over Penryn and be competitive at both stock and overclocked speeds. Only 45 days longer till we start seeing some review results, I think we'll be "wowed" again like with Conroe.

[Movieman]

gomeler;3294355']I'm certain Nehalem will offer higher IPC and efficiency over Penryn and be competitive at both stock and overclocked speeds. Only 45 days longer till we start seeing some review results, I think we'll be "wowed" again like with Conroe.

I don't know about that. Depends on what part of "this crowd' at Xs your addressing.
The gamers won't see huge gains but the encoding,photoshop,rendering, crunchers will so it all comes down to what do you do with your PC.
I don't game but if I did I'd buy a E8600, bang the biatch out to 4500 and call it a winner and bypass nehalem totally.
Now for what I do the question is what day are they delivering them so I can heist a truck and pass them out in the DC section!:rofl:

[junkmonk]

I'm going to jump straight on the nehalem bandwaggon. I've been in the market for a new system for the past 2 months, and I decided I'm going to wait for the new socket. Why? Because it's uncomfortable knowing that your system is obsolete. It sucks knowing that every day it just gets older and older, with no possible way to change that except to shell out for a brand new system.

For the sheer matter of upgradeability I'm going with nehalem. Whether it overclocks to 3.3 or 4.3 I will buy it, and bring it to its knees :D

There may be some little bugs due to the extensive amount of changes occuring in the next platform. But performance, WILL be better, there is no way that the Q9450 will beat the i7 920, even with overclocking... The added threads, new memory controller, and overall SIZE (1366vs775) will guarantee that this thing will be good.

I'm not scared of this thing being crap, it won't be!

[Epsilon84]

This chart pretty much validates the fact that enabling HT *decreases* performance, if you look at CCWS2K2. I remember there being quite a number of other applications where performance is significantly decreased with HT. Anand only shows one but there was a site that showed a lot more and actually cautioned against HT if those applications were to be used (and even showed how it BSOD'ed in some applications). Heck, do you really use 3dsMax or Divx encoding? If not, then it's hardly worth it, IMHO.

Yeah, one app shows a decrease, and that validates your point?! What about the other 25 apps that didn't show a decrease?

Well, the reality is that HT actually *decreases* performance on most apps that do not take advantage of multiple cores.

Well? :rolleyes:

Not to mention, the implementation of HT on Nehalem is different to the one on P4, and applications are a lot more multithreaded today than they were back in 2002...

[Donnie27]

Good luck with that, unless you're on the XE. I don't care what the five people on here buying XEs think, I'm not spending $999 on a CPU to get a decent overclock when I can get a multiplier-locked Penryn instead which will fall in price when these are released..

If the 2.66GHz version is able to reach 4GHz or close to it, then sure. Otherwise, there's no point, Penryn is better.

QFT! That's what I meant about the $999 processor, just look at past history?

I'm very impressed if that 2.66GHz model is anywhere near $300. All anyone has to do is look at what past models in that price shipped? More power to fellow Geeks looking forward to the XE and etc...... I'm happy for them:D Love to see them get every ounce of speed out of their processors. But just like watching a Drag race between very high performance cars. When it's over, I'll get in my Pick-em' up truck and go home.

[Nasgul]

$300..........but you know the first few months we're gonna be looking at $375-$400, I believe I paid $380 for me E6600 back in August 2006 :mad:

And I just bought a Quad Q6600 for $150...........never again I'm rushing to buy.

Though Quad Core and HTT along with it? :up:

[Donnie27]

$300..........but you know the first few months we're gonna be looking at $375-$400, I believe I paid $380 for me E6600 back in August 2006 :mad:

And I just bought a Quad Q6600 for $150...........never again I'm rushing to buy.

Though Quad Core and HTT along with it? :up:

I got mine before you and paid $327:up: The point I told the other poster not to mix up Intel tray price and VAR's sticking it to us! I didn't say they wouldn't screw us. Come on dewd, remember we laughed at Newegg selling
X6800 for $1,599:rofl::yepp:

[xVeinx]

Microcenter is THE place to get processors, they don't gouge :)

[T_Flight]

I'm going right down the middle. I'm not going low end, and I'm not going extreme. I'm going 2.93Ghz. 500 bucks for a CPu is not that bad. Now if I was buying CPU's every year it might be, but I don't.

The only reason I'm not going XE is becasue the price is *too* high. It is not worth 999 dollars. If it was priced at 700 dollars I would buy that thing and wouldn't think twice.

I'm gonna grab that 2.93 and run it for awhile, and after the system has ran stock for awhile, I'm gonna put a mild OC on it and see what it does...then I'm gonna go balls to the wall and bench a highpoint number. :yepp:

After that, I'm gonna go online with the new system, fire up folding and let it chew through about 8-9 thousand PPD. Then I'll stop F@H and let it simmer down for a bit and go up on a mission with my virtual pilot freinds, and destriy the DPRK in Falcon 4. Late this winter I also plan on installing the new CoD and ArmA, and own anybody that gets in front of me.

I'm certain this new Bloomfield will excel at all of those things, and I'll finally be able to do my video encoding and editing without becoming such a chore.

My rigs are not just gaming consoles. They are my "everything" rigs, and I build them for true multimedia, and multitasking.

BTW, the only crashes I have got that were related to HT is when I turned it off. Since then it has never been off, and there has never been a single error related to HT being tunred on. I've ran HT CPU's since their beginning.

I don't know where the notion came from that HT causes BSOD's, but I'd be willing to bet it came from AMD-Zone. :rolleyes:

[gallag]

My post wasn't directed at you... ;)



I don't think so...
http://www.anandtech.com/showdoc.aspx?i=1746&p=6

http://images.anandtech.com/reviews/cpu/intel/pentium4/3.06G/image002.gif

This chart pretty much invalidates the remainder of your post regarding HT...

This chart is from 2002, It is also on a completely different architecture.

[Epsilon84]

This chart is from 2002, It is also on a completely different architecture.

Yes, I realise that, why can't people take posts into context?! Is it really that hard?!

Look at what Bo_Fox posted, he was rambling on about how P4s w/HT ran slower in almost every non multi-threaded app with HT enabled which is absolute rubbish.

[Bo_Fox]

Well, that was my impression from a couple of years ago after seeing a few negative articles about HT. Many people were actually disabling HT, in fact (if you can remember)...

True, there are (and will be) lots more multithreaded apps that make good use of HT. If I do not hear of any compatibility/stability issues with HT enabled on the i7, then I might leave it enabled after all! Relax guys, I was just reacting to your "smart" posts in an equally "smart" manner, LOL.. it's sometimes like a fun game for me that I enjoy doing (debate could be addictive, huh?)..

[AuDioFreaK39]

Do any of you know when we will see top quality first-rate games (like Crysis) with SSE4 support? I have been anticipating this for quite some time now.

For Penryn and Nehalem owners, what would the real-world performance benefits be from implementing it? (assuming we don't see true multicore support for still some ways to come)

[xtclocker]

Wow I must buy one soon!

[Epsilon84]

Well, that was my impression from a couple of years ago after seeing a few negative articles about HT. Many people were actually disabling HT, in fact (if you can remember)...

I ran a Northwood based PC from 03 to 05, and I didn't run into any computability or stability problems at all. From experience, HT was definitely beneficial for general desktop usage, and I had it enabled at all times.

The only issues HT had were with certain server based workloads, which may be what you're referring to. But you acted as if you KNEW Nehalem with HT would reduce performance in most cases which is just ridiculous since you have nothing to back that up except bogus claims about performance issues on P4s, which I soon debunked.

[AuDioFreaK39]

by the way the Core i7 model names have been confirmed:

http://www.expreview.com/news/hard/2008-09-16/1221564352d10023.html

http://www.expreview.com/img/news/2008/09/16/intel-roadmap-september.png

[Shintai]

$300..........but you know the first few months we're gonna be looking at $375-$400, I believe I paid $380 for me E6600 back in August 2006 :mad:

And I just bought a Quad Q6600 for $150...........never again I'm rushing to buy.

Though Quad Core and HTT along with it? :up:

I payed about 320-330$ for mine in July 2006. You just need to shop a better place :yepp:

[T_Flight]

I hope that's a sign that NDA's are about to be lifted. Man, this wait is killing me, and I'm not gonna get caught up in X-mas price gouging.

If the rumors are right we are less than 2 weeks from release of this thing. Where are the numbers?

Another thing that is agravating me is the memory. I have no idea what to run on it, and haven't seen any tri-channel matched kits. I hope it's not gonna be a problem finding matched memory. I really wanna go with 3x2GB for a total of 6, but I really want it matched.

[Nosfer@tu]

Because folks will do just like they did with E6600 and NOT X6800 that got out numbered 10 to 1. A lot fewer folks even care about the $1000 dollar model. This is Xtreme Systems.org and many know higher overclocking room with lower end processors than those who buy the top models and do the same.

I was just saying that prices are up to the Distributors, OEMs and VAR, not just AMD or Intel. The price you mentioned wasn't an Intel price as you said. I didn't mean anything nasty or negative.



This isn't Xtremely Rich Folks.org either LOL! E6600 and Q6600 and now 9300, 9450 and 8400 seems to be the most used here. One of the aims of overclocking is adding Value by overclocking. Pay for a lowend model and end up with a higher end model.

Or am I wrong? But that wasn't the point. I still see folks giving Newegg's price and not Intel's Tray Price as I said.

Well Donnie27 Im not trying to disrespect anybody. I just fell the thread has taken a turn of "Damm I have to spend alot" instead of WOW the smallest version is CHEAP.

I'm really sick of folks talking about $1000 CPU's when 2.66GHz might cost about $300 or so. Or less than Conroe E6600 or Athlon X2 3800+. Get real folks what you guys want?Now yield must be danged good if 2.66GHz is the slowest they're going to ship Or has some one mentioned this already?

Thats what I quoted you for.

I dont understand why you get sick of that ?

Just enjoy your self because new technoligy is on its way :D

And so what if the top model costs 2-3 times more than the low end model.
it always has :)

Im going to place some smilies in the end here. So you and others know im not in any way looking for a figth :D

Im just debating :D:D:D :yepp::):):D:D;):up: :D

Be happy, be Nehalem :D

[Shintai]

I hope that's a sign that NDA's are about to be lifted. Man, this wait is killing me, and I'm not gonna get caught up in X-mas price gouging.

If the rumors are right we are less than 2 weeks from release of this thing. Where are the numbers?

Another thing that is agravating me is the memory. I have no idea what to run on it, and haven't seen any tri-channel matched kits. I hope it's not gonna be a problem finding matched memory. I really wanna go with 3x2GB for a total of 6, but I really want it matched.

"Matched" memory is a joke. Just buy 3 2 packs of the same thing if you want trichannel and 6 sticks.

[informal]

Good news for Nehalem is the new low voltage DDR3 Jedec standard.It could mean OCing the ram will be easier with the new low voltage sticks than the current ones.Any information when will the first low v. sticks be available on the market?

[AuDioFreaK39]

Good news for Nehalem is the new low voltage DDR3 Jedec standard.It could mean OCing the ram will be easier with the new low voltage sticks than the current ones.Any information when will the first low v. sticks be available on the market?

just in - Bloomfield officially supports only DDR3 800 and 1066, not 1333 (http://www.fudzilla.com/index.php?option=com_content&task=view&id=9475&Itemid=1)


How could that be good :(

[informal]

Well the official support is one thing,user OCing is another.I can bet that the limitation is just a formal one.You will be able to clock your ram higher,but it will depend from board to board(like reported previously).Also of interest is the ref. clocking and how it affects the ram speed.It's still early to say how the officially supported ram clocks will affect the overclocking crowd.

[Kingcarcas]

Now for what I do the question is what day are they delivering them so I can heist a truck and pass them out in the DC section!:rofl:
Count me in :up: Curious what kinda PPD 2/4 instances can pump out with the SMP client.

[Hornet331]

by the way the Core i7 model names have been confirmed:

http://www.expreview.com/news/hard/2008-09-16/1221564352d10023.html

http://www.expreview.com/img/news/2008/09/16/intel-roadmap-september.png

ha, that roadmap is promising for getting a i7-965, no other top model to replace it for 3 quaters and loose value like mad, just what happend with the QX9650. :cool:
I think i get rid of mine in oct to get at least ~550€ (from 980€ in dez last year :p:) out of it, buy a E2160 in the meantime and then jump on the 965 asap.

[Shintai]

ha, that roadmap is promising for getting a i7-965, no other top model to replace it for 3 quaters and loose value like mad, just what happend with the QX9650. :cool:
I think i get rid of mine in oct to get at least ~550€ (from 980€ in dez last year :p:) out of it, buy a E2160 in the meantime and then jump on the 965 asap.

Thats not true. It says 965 _or_ higher :)

[Hornet331]

Thats not true. It says 965 _or_ higher :)

well, in Q408 and Q109 thers only the 965 alone and then in Q3 thers "or higher", i dont expect thers a higher clocked Ci7 in Q2 so this 3 quarter are quite reasonable. ;)

[Slovnaft]

guys the ongoing NDA is starting to scare me.
I haven't seen many meaningful data regarding nehalem other than intel-spawned misinformation and the one review site that did some video encoding (whuptydoo?), and there's certainly been nothing regarding QPI from an overclocking headroom standpoint.
My spideysenses tell me core i7 might be a huge let down especially in its infant stages, and worst of all for those not interested in dropping a grand on a silicon slab.
Maybe I'm just trying to rationalize my recent purchases in penryn, but I fear nehalem won't clock nearly as well as these chips have. I mean, what if Intel really is going the way of AMD with its barely OCable HT interface?
am I the only one experiencing these P4 HT pangs of wariness? is everyone else just creaming their jeans to get ahold of these chips and if so could you provide me some reasons for excitement?

[Hornet331]

guys the ongoing NDA is starting to scare me.
I haven't seen many meaningful data regarding nehalem other than intel-spawned misinformation and the one review site that did some video encoding (whuptydoo?), and there's certainly been nothing regarding QPI from an overclocking headroom standpoint.
My spideysenses tell me core i7 might be a huge let down especially in its infant stages, and worst of all for those not interested in dropping a grand on a silicon slab.
Maybe I'm just trying to rationalize my recent purchases in penryn, but I fear nehalem won't clock nearly as well as these chips have. I mean, what if Intel really is going the way of AMD with its barely OCable HT interface?
am I the only one experiencing these P4 HT pangs of wariness? is everyone else just creaming their jeans to get ahold of these chips and if so could you provide me some reasons for excitement?

No Data?

Well i guess you need to look a bit harder, ther are reveiws from Anad,Toms,Hexus + here some user ran benches, thats more we got then with the launch of core. :yepp:

[Jacky]

Anandtech, tom's, hexus, jcornell (sp?) all benchmarked nehalem, but there was not much overclocking involved. If anything nehalem launch is much smoother and faster than penryn, or am I wrong?

[Slovnaft]

I was more pointing out the problem of interpretation and incompletion.
Anandtech's review is hard to calibrate because of the odd benches, though the numbers seem indimidating i guess, tom's is incredibly incomplete--no more than a cpuz validation and a brief overview of what's there and the same goes for hexus.
I guess I should be happy that everyone seems to be giving the blanket statement "these chips are awesome" but that makes me wary.

[Hornet331]

I was more pointing out the problem of interpretation and incompletion.
Anandtech's review is hard to calibrate because of the odd benches, though the numbers seem indimidating i guess, tom's is incredibly incomplete--no more than a cpuz validation and a brief overview of what's there and the same goes for hexus.
I guess I should be happy that everyone seems to be giving the blanket statement "these chips are awesome" but that makes me wary.

lol this are previews, you dont run a full benchmark course in previews, just a few selevted. If you want a full review you have to wait till t's released.

Hexus had quite a few numbers, but pulled it a view hours after it was released. Afaik they had the full sandra suite benched, hexus pi fast, games (ET, lost planet), pi, cinebench, and some stuff i forgot. :yepp:

Also toms clocked nehalem to 4ghz, which shows that even not so skilled ocers can clock nehalem, on the other hand whan i comapre this to the initial agane overclocking...

[Slovnaft]

where is this OC? I didnt see that anywhere...

[Shintai]

where is this OC? I didnt see that anywhere...

http://www.google.co.uk/search?hl=en&q=nehalem+4ghz&meta=

[Donnie27]

Well Donnie27 Im not trying to disrespect anybody. I just fell the thread has taken a turn of "Damm I have to spend alot" instead of WOW the smallest version is CHEAP.



Thats what I quoted you for.

I dont understand why you get sick of that ?

Just enjoy your self because new technoligy is on its way :D

And so what if the top model costs 2-3 times more than the low end model.
it always has :)

Im going to place some smilies in the end here. So you and others know im not in any way looking for a figth :D

Im just debating :D:D:D :yepp::):):D:D;):up: :D

Be happy, be Nehalem :D

Simple.

Post #3


Not as bad as it could be.

Normaly they are released at 1100-1250 If I remember correct.


They'll be released at $999 because Intel releases them. NOt 1150 to 1250. I simply tried to explain that Intel is not responsible for folks like Newegg and others jacking the Prices up higher after the fact. If you made that point, I'm not even posting to you. If post #3 said;

Not as bad as it could be.

Normaly they are released at $999 If I remember correct.


My reply is; "Yes, it has been stated as such.

No need for :rofl: or :rolleyes: or :ROTF:

[Slovnaft]

http://www.google.co.uk/search?hl=en&q=nehalem+4ghz&meta=

Well...ok...

It seems the original snapshot has been removed, but there are plenty of paragraph-shaped blurbs about it.
I was about to eat my shoe when I saw that vcore was pushed to 1.576 to achieve this. Now, I'm not exactly sure what this means for Nehalem's arch., but I know that 1.45-1.5 just to bench was enough to kill my Q9450 after ~10hrs. And, since Nehalem runs on such low stock voltage, the assumption would be that it would tolerate lower max vcore than yorkfield. Now, this is just an assumption and I could be wrong, the new architecture could be incredibly tolerant of voltage, but it seems to me that 1.576vCore might be enough to kill a chip PDQ, especially under air.
I mean, I'm sure I couldve gotted 4.11 with my Q9450 with 1.576v, I was pushing 4080 at 1.45...

So it's mixed news imo. but i guess I'll shut my mouth until it comes out and we get some meaningful data.

[Hornet331]

^

just bad luck, i ran my QX9650 @ 1,66V for several hours for benching (4500mhz) and its still healthy and kicking, on my main ocing forum there are even guys that push 1,7V+ and the cores are still working without any problem.

Plus i think you put way to much voltage for 4ghz, cause most yorkfield can reach 4ghz with sub 1,4V stable. I run now 3,5ghz with 1,264V. But anything above 4GHz need insane voltages.

[Slovnaft]

well it needed high voltage because it was really straining its fsb limitations at 510mhz.
I think it was when i got a little crazy and went to 1.55v to try for 520fsb many times that did it in.
Anyways, sovery OT.
Bottom line, I would certainly not consider a 45nm at 1.576 to be safe and 'stable'

[AuDioFreaK39]

I was going through Intel's August 20, 2008 Roadmap and I noticed the following slide:

http://download.intel.com/pressroom/kits/events/idffall_2008/SSmith_briefing_roadmap.pdf

http://img530.imageshack.us/img530/5687/corei7extreme965slideqo2.jpg

"Extreme SKU has overprotection removed for overclocking"


Does this mean the 965 will see even further overclocking advantages from just having an unlocked multiplier?

[oohms]

"Extreme SKU has overprotection removed for overclocking"

That's like saying "I just the whole bottle!"

Hopefully they don't neuter the non XE chips too much

[Slovnaft]

Does this mean the 965 will see even further overclocking advantages from just having an unlocked multiplier?

Well, any chip will, the question is whether these chips will OC well without an unlocked multi, essentially, how flexible QPI is.
I mean, that impacts the extreme chips as well.

[Donnie27]

Well, any chip will, the question is whether these chips will OC well without an unlocked multi, essentially, how flexible QPI is.
I mean, that impacts the extreme chips as well.

Aren't Cores and QPI ran independently of eachother?

[Slovnaft]

No, QPI is on-die.
take a look at the architecture:

http://chip-architect.com/news/Nehalem_at_1st_glance_.jpg

[Bo_Fox]

I ran a Northwood based PC from 03 to 05, and I didn't run into any computability or stability problems at all. From experience, HT was definitely beneficial for general desktop usage, and I had it enabled at all times.

The only issues HT had were with certain server based workloads, which may be what you're referring to. But you acted as if you KNEW Nehalem with HT would reduce performance in most cases which is just ridiculous since you have nothing to back that up except bogus claims about performance issues on P4s, which I soon debunked.

Hmmm.. a flawed interpretation of what I acted like would get you "pwned" in a frag-fest in an instant. "Soon" is not quick enough.

Seriously though, (since memory usually serves me correctly), there were quite a few articles on HT that it is not always desirable. If you are a "know-it-all", you would tell me the percentage of owners of such chips who disabled HT.

[Hornet331]

Hmmm.. a flawed interpretation of what I acted like would get you "pwned" in a frag-fest in an instant. "Soon" is not quick enough.

Seriously though, (since memory usually serves me correctly), there were quite a few articles on HT that it is not always desirable. If you are a "know-it-all", you would tell me the percentage of owners of such chips who disabled HT.

less then one percent. :rofl:

Since 90% of the useres dont even know what a bios was, or how to enter. Plus many of the enthusiats also ran HT, maybe disabled it for ocing, but for daily usage they enabled it.

[Shintai]

HT gave one thing that was priceless on the P4. Nomatter if performance increased or dropped alittle. Creamy Smoothness(TM)
Dualcore smoothness without having one.

[R101]

Yep, and at a time, if you did some CPU consuming tasks in the background there was no alternative. A64's were quick,
but no way you could do office work while rendering on them without having hiccups.
X2's changed that, but it was much much later.

[Sesto Sento]

Well...ok...

It seems the original snapshot has been removed, but there are plenty of paragraph-shaped blurbs about it.
I was about to eat my shoe when I saw that vcore was pushed to 1.576 to achieve this. Now, I'm not exactly sure what this means for Nehalem's arch., but I know that 1.45-1.5 just to bench was enough to kill my Q9450 after ~10hrs. And, since Nehalem runs on such low stock voltage, the assumption would be that it would tolerate lower max vcore than yorkfield. Now, this is just an assumption and I could be wrong, the new architecture could be incredibly tolerant of voltage, but it seems to me that 1.576vCore might be enough to kill a chip PDQ, especially under air.
I mean, I'm sure I couldve gotted 4.11 with my Q9450 with 1.576v, I was pushing 4080 at 1.45...

So it's mixed news imo. but i guess I'll shut my mouth until it comes out and we get some meaningful data.


Strange that your Yorkfield died with such a low vcore...

I know it's not the same, but i've been runing my E8400 at 4320mhz (9x480) @ 1.475v (cheap Watercooling) for 11 months straight (24/7) now and still rock solid as if it was new...

[Epsilon84]

Hmmm.. a flawed interpretation of what I acted like would get you "pwned" in a frag-fest in an instant. "Soon" is not quick enough.

Seriously though, (since memory usually serves me correctly), there were quite a few articles on HT that it is not always desirable. If you are a "know-it-all", you would tell me the percentage of owners of such chips who disabled HT.

Sorry mate, I'm not the one who definitively stated that almost every single threaded app loses performance with HT on. You still haven't provided any evidence to prove that point, and I don't think you will anytime soon. ;)

The majority of articles on HT were more favourable than not, most focused on the improvement in general responsiveness during multitasking in the age before dual core CPUs.

If you could just provide a link to an article that shows HT having massive performance and stability issues as you claim, then I would debate it on those merits, but so far all you've been sprouting is hot air.

[Slovnaft]

Strange that your Yorkfield died with such a low vcore...

I know it's not the same, but i've been runing my E8400 at 4320mhz (9x480) @ 1.475v (cheap Watercooling) for 11 months straight (24/7) now and still rock solid as if it was new...

Well in all fairness i did push it up to 1.55 for a few hours to make various 3d bench attempts, i think once i mightve even tried 1.6. I really wasn't too careful.
And this is all on air.


And back to nehalem, this is the first single die quad core isn't it?
all else aside i think nehalem is a miracle of microarchitecture.

[Sesto Sento]

yes indeed, it's the first single die quad core from intel

[informal]

Well in all fairness i did push it up to 1.55 for a few hours to make various 3d bench attempts, i think once i mightve even tried 1.6. I really wasn't too careful.
And this is all on air.


And back to nehalem, this is the first single die quad core isn't it?
all else aside i think nehalem is a miracle of microarchitecture.

1.6V for a 45nm chip with high-K/mg is a bit of a stretch.

Nope,Nehalem is not a first single die quad core(generally speaking).As for a miracle,it's quite impressive and extremely complicated.That's why intel decided to make it @ 45nm and not @ 65nm.It's base line Penryn ,but with a lot of both small and big tweaks.The biggest change is ,for sure, the riddance of FSB.QPI+ IMC is going to help them a lot in multi socket server market.On desktop,i doubt it make a big difference(apart from SMT and IMC that will make a difference compared to C2Q).

[Slovnaft]

But more than performance differences, i guess i was trying to get at intel's trend of bringing resources into on-die architecture. Between nehalem's IMC and whatever this Larabee BS turns out to be, it seems like intel is trying to bring as many tasks as possible onto low-latency structures on the CPU.
While in theory this seems like a somewhat logical progression, i guess nehalem and larabee will be our litmus test of whether it's "too soon". i think it's definitely to soon for on-die graphics engines.

[Donnie27]

1.6V for a 45nm chip with high-K/mg is a bit of a stretch.

Nope,Nehalem is not a first single die quad core(generally speaking).As for a miracle,it's quite impressive and extremely complicated.That's why intel decided to make it @ 45nm and not @ 65nm.It's base line Penryn ,but with a lot of both small and big tweaks.The biggest change is ,for sure, the riddance of FSB.QPI+ IMC is going to help them a lot in multi socket server market.On desktop,i doubt it make a big difference(apart from SMT and IMC that will make a difference compared to C2Q).

I agree but it seems some Geeks want to have it both ways. I'd said and been jumped by some for saying FSB and IMC on the Desktop isn't that big of a deal. I said then that Nehalem gains will be had from Core improvements, not just QPI and IMC that's meant for servers and 4 plus sockets, since Dual Socket Boards already use 2 FSB's with great results. Ask Movieman?:up: Where QPI and SMT will help on the desktop is VT and Multitasking.

[mascaras]

Core i7 overclock feature changed

There’s always a demand for overclock, not only the elite overclocking community but also mobo vendors want Intel to change the plan, not limiting Core i7’s OCbility any more.

So Intel changed ideas recently, going to unlock memory ratios adjustments on Core i7 940 and 920. Before September Intel want this feature only exist in Core i7-965 XE.

So what can we expect with a Core i7-940/920? Memory will not only running under 800/1066, you can tune up for higher frequency, because Core i7 940/920 now have the ability to change memory multipliers, so mobo maker can write a BIOS with option to increase memory frequency. Also, changes memory multipliers can also allows QPI speed to be changed, but documents pointed out that raise QPI speed will only gain little performance.



http://en.expreview.com/2008/09/19/core-i7-overclock-feature-changed/#more-927





regards

[xVeinx]

I was under the impression that the multi-lock was hardware enabled, so how can they make a change like this so close to launch? Granted, they could probably find an outlet for those with locked multis, but even so.

[STaRGaZeR]

I was under the impression that the multi-lock was hardware enabled, so how can they make a change like this so close to launch? Granted, they could probably find an outlet for those with locked multis, but even so.

If I got it right they're unlocking the memory ratios (dividers?), not the core multi.

[Donnie27]

If I got it right they're unlocking the memory ratios (dividers?), not the core multi.

That's what it looked like to me too!

[informal]

Even though this is a nice feature enabling higher memory clock speeds ,this still doesn't solve the "140MHz wall" with ref. QPI clock on locked(non XE) chips..Why intel doesn't unlock the multi up to a few steps, ie. 21,22,23 for the lowest model?This way you could at least hit 3.2GHz with 23x multi and 140Mhz ref clock which could be enough to a lot of people..

[Loque]

to wait for i7 for gaming or buy a Q9550 ... that's the question..

[Blauhung]

Even though this is a nice feature enabling higher memory clock speeds ,this still doesn't solve the "140MHz wall" with ref. QPI clock on locked(non XE) chips..Why intel doesn't unlock the multi up to a few steps, ie. 21,22,23 for the lowest model?This way you could at least hit 3.2GHz with 23x multi and 140Mhz ref clock which could be enough to a lot of people..

I forget, where did 140MHz come up as a wall?

And by just manipulating the turbo settings you can get the multi to stay at 22 permanently.

nevermind, found it.
http://www.xtremesystems.org/forums/showpost.php?p=3263689&postcount=1196
My first impressions with own Nehalem ...

1. BIOSes and whole platform is still very immature ...
2. Only little raised reference clock causes massive unstability ...
3. Ocing of locked-multi Nehalems will be a problem ... Reference clock of Bus is not like a FSB - my max stable clock is 140MHz from 133MHz :down:
4. If you want great clocking, you need a Extreme chip with unlocked multi
5. With these early BIOSes is Memory performance poor (or Everest dont know how to measure it?)
6. Power consuption is in idle with all power savings great, but in fully load is far away from 45nm Quads ...
7. Max stable ocing was on my chip about 3500MHz ...
8. Performance is various, read my previous posts in this thread ...
9. We have two months to launch, and i believe Intel will do many improvements to that date. Will see next month, with new BIOSes and maybe different mobo ...

I wouldn't trust the 140 value just yet. He was using a very early stepping of both Bloomfield and Tylersburg along with very immature BIOS.

[Blauhung]

I was under the impression that the multi-lock was hardware enabled, so how can they make a change like this so close to launch? Granted, they could probably find an outlet for those with locked multis, but even so.

I'm unsure if the lock is package level or if this is one of the settings that get fused on die at sort. But I am under the impression that this was probably a decision made a bit ago that just hadn't hit the news yet.

But could be wrong

[GAR]

to wait for i7 for gaming or buy a Q9550 ... that's the question..

I wouldnt upgrade to a new platform just yet, wait about 6 months to let it mature a bit, it should be a good platform once it matures, lets see if AMD has an answer.

[Hornet331]

I wouldnt upgrade to a new platform just yet, wait about 6 months to let it mature a bit, it should be a good platform once it matures, lets see if AMD has an answer.

none we know off. :p:

Amds roadmaps are pretty much clear till the end of 2009 and i dont see anything that would "threat" nehalem.

[GAR]

none we know off. :p:

Amds roadmaps are pretty much clear till the end of 2009 and i dont see anything that would "threat" nehalem.

Seems like another year dominated by INTEL

[Nasgul]

Seems like another year dominated by INTEL

20 out of 23 previously...........and now 3 for 3.

The one thing I don't get though, with the introduction to triple channel? Are G.Skill and the rest going to start shipping triple channel kits or are we going to be stuck buying two 4GB (2x2) kit of DDR3?

And how about having to move to 64-bit? With 6GB of DDR2 (2GB x 3), seems like it'd be inevitable.

We're forced (somewhat) to do a major upgrade overall, not just CPU.

[mikeyakame]

http://en.expreview.com/2008/09/19/core-i7-overclock-feature-changed/

looks like the non XE chips just got even better if this is in fact true!

[@@@@]

Going to a nehalem CPU means a whole platform upgrade but if you are already in DDR3 modules with the current motherboards I think you only need to change motherboard + CPU and plug in the rest from you previous system but a lot of the quality motherboards not necessarily top of the range are expensive. I would take GAR's advice wait for the whole platform to mature not only the CPU but also consider the motherboards and RAM

[Shintai]

20 out of 23 previously...........and now 3 for 3.

The one thing I don't get though, with the introduction to triple channel? Are G.Skill and the rest going to start shipping triple channel kits or are we going to be stuck buying two 4GB (2x2) kit of DDR3?

And how about having to move to 64-bit? With 6GB of DDR2 (2GB x 3), seems like it'd be inevitable.

We're forced (somewhat) to do a major upgrade overall, not just CPU.

You are not forced to do anything. And if you aint already at 64bit OS...its really about time!!

And you could also just run 3x1GB if you feel thats the fun of the day :p:

Matched memory is also a joke. Just get 3 sticks...but again. It will also work with both single and dualchannel.

[Bo_Fox]

less then one percent. :rofl:

Since 90% of the useres dont even know what a bios was, or how to enter. Plus many of the enthusiats also ran HT, maybe disabled it for ocing, but for daily usage they enabled it.


:rofl::clap: You just pwned me! LOL!



Sorry mate, I'm not the one who definitively stated that almost every single threaded app loses performance with HT on. You still haven't provided any evidence to prove that point, and I don't think you will anytime soon. ;)

The majority of articles on HT were more favourable than not, most focused on the improvement in general responsiveness during multitasking in the age before dual core CPUs.

If you could just provide a link to an article that shows HT having massive performance and stability issues as you claim, then I would debate it on those merits, but so far all you've been sprouting is hot air.

Hot air? You mean BS? :p: I never said "massive"--you gotta stop making mountains out of mole-hills. It's just that when I were reading all those articles, there were "quite a few" articles that were not so positive about HT for general usage. The same thing applied for NCQ (Native Command Queuing) with hard drives. I guess as we adopt the new features, more and more applications are optimized to take advantage of them. I do not want to "obey" your demands by digging up for a link. Perhaps you could just argue back instead?



HT gave one thing that was priceless on the P4. Nomatter if performance increased or dropped alittle. Creamy Smoothness(TM)
Dualcore smoothness without having one.

Whoa! Then why didnt those articles say it like you did, when HT was new? Pentium 4 was hardly helped by HT against the massively successfuly A64. After all, HT is nowhere as smooth as REAL MULTI-CORES (TM). HT might give like 10% improvement, but not 100% improvement scaling like dual cores.

Debate.. debate.. let's just drop it. I'll most likely leave this thread.

[Shintai]

Whoa! Then why didnt those articles say it like you did, when HT was new? Pentium 4 was hardly helped by HT against the massively successfuly A64. After all, HT is nowhere as smooth as REAL MULTI-CORES (TM). HT might give like 10% improvement, but not 100% improvement scaling like dual cores.

Debate.. debate.. let's just drop it. I'll most likely leave this thread.

:rofl:

Nobody said 100%, and you fail again to show any of your "statements".

http://www.2cpu.com/articles/42_4.html

HT worked, HT made it faster, HT made it smoother. You dont need a fulldualcore to make windows feels smoother. When your its just a 0-1% task that needs to run while another talls.

You gave me a good laugh tho :ROTF::ROTF::ROTF:

[Bo_Fox]

:rofl:

Nobody said 100%, and you fail again to show any of your "statements".

http://www.2cpu.com/articles/42_4.html

HT worked, HT made it faster, HT made it smoother. You dont need a fulldualcore to make windows feels smoother. When your its just a 0-1% task that needs to run while another talls.

You gave me a good laugh tho :ROTF::ROTF::ROTF:

You failed to answer me why it didnt help the P4 against A64 single cores (before the X2's were finally released much later). :ROTF::ROTF::ROTF:

[Shintai]

You failed to answer me why it didnt help the P4 against A64 single cores (before the X2's were finally released much later). :ROTF::ROTF::ROTF:

Because games was and to a certain degree still is singlethreaded. But to say HT was a failure or didnt have a bonus is pretty retarded. And where is your proof?

[FischOderAal]

I'm with Shintai on this one. HT definately brought some benefits, question is if the main reason was the architectual "failure" of Netburst. Atom benefits as well, but this is due to the In-Order-Architecture. The new architecture on the other hand doesn't seem to benefit much from HT (or SMT), as Intel abandoned that feature beginning from Banias. Banias was a mobile-chip though.

[mikeyakame]

perhaps HT on netburst m-arch was simply born before its era. its era is now, not 5 or so years ago

[Nasgul]

You failed to answer me why it didnt help the P4 against A64 single cores (before the X2's were finally released much later).
Back in those days, the only success A64 had? were because everyone had only one word that mattered the most: "gaming", and thanks to that, A64s were supposedly better.

Except P4s were faster than A64s at multi-media, content creation, office apps and anything and everything that was not "gaming", so despite of all of that? We all knew that Intel still had the best "gaming" CPU, Dothan.

To this date, "gaming" is such, if not perhaps the most, an important part for buying a PC.

[mikeyakame]

Hmnm why must people get so heated up over the past...we live in today, yesterday is a memory and tomorrow could bring anything. live for tomorrow not yesterday.

[Metroid]

Going to a nehalem CPU means a whole platform upgrade but if you are already in DDR3 modules with the current motherboards I think you only need to change motherboard + CPU and plug in the rest from you previous system but a lot of the quality motherboards not necessarily top of the range are expensive. I would take GAR's advice wait for the whole platform to mature not only the CPU but also consider the motherboards and RAM

I always thought that if we were to go for DDR3 using CD2 would be pretty pointless since DDR3 prices were/are still high and it is tad better compared to DDR2.

So since most of us knew dual channel would become extinct by the fact Intel did create a platform which wanted to show to everyone an intense memory gains regarding the only clearly seen weakness of its current platform 775 and its derivatives. It was very clear that only a matter of time would hold its use until tri-channel would be deployed in the new platform.

So The reality is, a tri-channel memory at 1066 will be much better than 2 DDR3 dual-channel at 1600, if the circumstances are given. So who used DDR3 in the previous or current platform lost a considerable amount of money because it might not be of any use in the next new platform.

Metroid.

[JumpingJack]

You failed to answer me why it didnt help the P4 against A64 single cores (before the X2's were finally released much later). :ROTF::ROTF::ROTF:

SMT is a threading technique, specifically simultaneous multithreading. This is not a novel or new concept, not even for Intel when they implemented in P4. Several microprocessor designs have utilized the concept to improve performance, the catch is that -- as the name implies -- it will only be useful for mulithreaded situations.

Games are hard to multithread, and back in the day, no game was multithreaded. Add to that that the P4 basic architecture was botched toward the end, gave the A64 a good lead. P4's really sucked at gaiming, though, because of the deep pipeline -- where branch mispredictions incurred high penalties by stalling the processor as much as 30 cycles or more and game code is very branchy.

http://www.firingsquad.com/hardware/quake_4_dual-core_performance/page3.asp

Shortly after HT came on the scene, Intel worked with iD to produced a multithreaded patch for quake 4 ... in this case, it should be obvious that HT does indeed help -- but only in cases where the code is multithreaded.

[Bo_Fox]

Because games was and to a certain degree still is singlethreaded. But to say HT was a failure or didnt have a bonus is pretty retarded. And where is your proof?

Good answer! That was a lame question anyways..

Alrite, I'll post some stuff.. they're just a quick google search that should include some of what I read on HT from a while ago when it was new and being discussed all over the 'net...


Hyperthreading hurts server performance, say developers
Intel's Hyperthreading Technology (HT) is being blamed for server performance problems.
http://news.zdnet.co.uk/hardware/0,1000000091,39237341,00.htm

Intel HyperThreading: Problems You Should Know
"there are no clear advantages the HyperThreading can offer us now" ---i clearly remember reading this statement
http://www.xbitlabs.com/news/cpu/display/news168.html

Hyper-Threading Considered Harmful
http://www.daemonology.net/hyperthreading-considered-harmful/

John Goodacre, the writer of the report, went as far as to say that SMT was only really useful as a way to rescue a modicum of performance from when you have a mismatch between the processor's speed and the rate at which you can feed it something to crunch on. Fun to read, at least.. seeing how it states HT not being worth it when it comes to power savings and real physical cores.
http://www.theinquirer.net/en/inquirer/news/2006/08/02/arm-is-no-fan-of-hyperthreading