NextGen VS. CurrentGen CPU: 3DMark06

here comes 3Dmark06:

http://www.expreview.com/img/news/071106/3dmark06.png

Quote:

---

From the final result, Phenom only lose QX6850 about 5.5%. But the CPU Score is losing 9.11% . If the rival change to QX9650, the gap will even bigger. If we compare to other subitem, that dual core E6850 will boost Graphic system better than Phenom X4. Comparing to Phenom X4 and E6850, All the advantage score are coming from the CPU score. Phenom X4 does bad in HDR and SM subitem tests.

In fact, LOTs of games are are not have the enough dual core or quad core optimize than 3Dmark06. In other test, you may find out even a dual core Core 2 is better than Phenom X4. That means not enough optimize support for our Phenom X4.

---

http://www.expreview.com/news/hard/2...85d6811_1.html

Why you can't run Phenom with normal ram speeds and timings?

Quote:

---

Originally Posted by Kasparz

Why you can't run Phenom with normal ram speeds and timings?

---

Maybe divider issues since they run Phenom at 3.0GHz and that is definetly overclocking it (like 20%+)

huh?
http://www.xtremesystems.org/forums/...4&postcount=80
Why he used 7 tRP value then? Its like umm...downclocking? Is there any problem to tighten tRP?

ive been looking at these and what is the chipset performance difference and what about vista were u can actually use the entire quad at teh same time efficiently


but good post space man

I think he mentioned in another thread that CPU-Z was misreporting the timings even after he set 5-5-5-18 in BIOS. :down:

Quote:

---

Originally Posted by Kasparz

huh?
http://www.xtremesystems.org/forums/...4&postcount=80
Why he used 7 tRP value then? Its like umm...downclocking? Is there any problem to tighten tRP?

---

CPU-Z problem,All setting is by SPD,means to be 5-5-5-18.

just wait CPU-Z update.

Quote:

---

Originally Posted by metro.cl

Maybe divider issues since they run Phenom at 3.0GHz and that is definetly overclocking it (like 20%+)

---

yes. the Phenom was oc on 200MHzX15,and the RAM frequency would be:

3000/7=428.57MHz or 3000/8=375MHz.

It does seem slightly wrong to dissadvantage AMD by running slower Ram than the Intel, its not the winning difference, but still. Same as the Crysis benchmark.

Quote:

---

Originally Posted by The0men

It does seem slightly wrong to dissadvantage AMD by running slower Ram than the Intel, its not the winning difference, but still. Same as the Crysis benchmark.

---

It's a divider issue @ 3GHz, either the RAM will run faster or slower than the Intel system.

I actually would rather Expreview run the RAM @ 3000/7=428.57MHz, who cares if it's slightly faster than on the Intel system, at least we won't get all the whining and crap about AMD being disadvantaged (not a slur at you, just what I've noticed from the Crysis threads).

Then again, we'll probably get Intel fanboys chiming in instead that Intel is then disadvantaged, even if Intel wins the benchmark. LOL :ROTF:

Quote:

---

Originally Posted by metro.cl

Maybe divider issues since they run Phenom at 3.0GHz and that is definetly overclocking it (like 20%+)

---

It is precisely the memory divider issues, any non-even CPU multiplier will down clock the memory at stock system clock speeds (200 MHz). Don't you feel cheated :)

This is a well known phenomena (pardon the pun) of AMD's memory controller, most AM2 reviews made note of this:

Example: http://www.anandtech.com/cpuchipsets...px?i=2762&p=10

For a multiplier of 15, the fastest a Phenom will clock DDR2-800 will be 375 Mhz not 400 Mhz.... so what you see is what you are gonna get. To get 400 Mhz on the ram, you will need to force the DDR2-800 ram to 428 (i.e. 3000/7 = 428 and thus risk stability) or buy DDR2-1067 and force it do DDR2-800 (i.e. 3000/7 = 428 running psuedo DDR2-856), but then you would probably want to multiply that memory up to to closest to 1067 as you can get, but it too will underclock it's full rating.

Everyone should note -- if you purchase an AMD CPU with a non-even CPU multiplier, the memory will be down clocked from it's best possible... this is just the way the AMD memory controller works.

The short answer -- the memory speed reported above is exactly where you would expect an AMD CPU to run DDR2-800 memory with a CPU multiplier of 15...

Jack

Quote:

---

Originally Posted by expreview

yes. the Phenom was oc on 200MHzX15,and the RAM frequency would be:

3000/7=428.57MHz or 3000/8=375MHz.

---

Is 3000/6=500MHz with 1066 DDR2 not possible?
Why is the northbridge running at 16000MHz, like on a non split power board?
ES issue?

Quote:

---

Originally Posted by The0men

It does seem slightly wrong to dissadvantage AMD by running slower Ram than the Intel, its not the winning difference, but still. Same as the Crysis benchmark.

---

idd!

i'd like to see another test: downclock the intels to be on par with the phenom @stock, not the other way around.

Quote:

---

Originally Posted by RaZz!

idd!

i'd like to see another test: downclock the intels to be on par with the phenom @stock, not the other way around.

---

What difference would that make? Still clock for clock.

I think he wanna see scaling capabilities.

Quote:

---

Originally Posted by AbelJemka

I think he wanna see scaling capabilities.

---

Oh yeah, according to some people, K10 introduces superlinear scaling... +10% clock = >+10% performance :ROTF: :clap:

Well I'm very impressed with Yorkfield...stock clocks, highest score of the bunch, miserly power draw.

Don't forget Yorkfield easily oc to 4ghz+ stable on air :)

Now those are the resulst i was waiting for :)

But what's teh platform?
mobo wich chipset?

Quote:

---

Originally Posted by BeardyMan

Now those are the resulst i was waiting for :)

But what's teh platform?
mobo wich chipset?

---

RD790 dude... open the link. ;)

Quote:

---

Originally Posted by Epsilon84

RD790 dude... open the link. ;)

---

if i only could :rolleyes:

memory speed for k10 no longer calculate with cpu clock. wait for new version of cpuz or you can use everest to show memory speed correctly..

grrr stop teasing!!! i want a phenom!!!!

Quote:

---

Originally Posted by justapost

Why is the northbridge running at 16000MHz, like on a non split power board?
ES issue?

---

i was thinking same thing

It seems a long time ago now that the summer was about to start and the K10 was 40% faster than Intel. I had the impression that it was not going to be 40%, more like 15%-20% and the main question was whether it would be able to overclock enough, not that it would lose in IPC.

Now it seems that for IPC it is about level or slightly behind ( with current testing setups ) and therefore not enough to offset any disadvantage with overclocking. The prices seem OK though. Disappointing it has to be said. AMD would have been better never to mention the 40%.

The main problem I see is that AMD have only got a die shrink coming up to try and rescue the situation, and historically die shrinks have never been that good for them. Meanwhile Intel is going to be throwing an integrated memory controller and their new bus ( whatever it is called this week! :) ) which should give a large boost, if AMD's K8 is anything to go by over K7. And we know the Intel 45nm process is good now and will be even better by Q3 2008.

AMD really needed to get the lead back at this point for Intel to pinch it back and then AMD to ..etc etc. That's not going to happen it seems.

Regards

Andy

i don't believe that this runs on a RD790 board :down:

just look at the NB frequency, do they use ganged or unganged IMC mode? (che k kyosens thread for more info on this) , their ram freq is weird (K10 doesn't derive mem clock from core speed like itt used to be on K8), the timings are not OK.

That 1600MHz HT frequency is I think half of what it should be. It's definitely low and the board either doesn't support split power planes or has no new BIOS support for that. Either way, it's nothing too short of what I expected of this arch by September. None of those CPUs impress me but the new Q9450 and Q9550.

NB freq shoudl be 2.6ghz and not 1.6ghz , yes it makes a difference, whereas HTT at 266 or 200 doesn't.

Again way to go expreview with this half complete review :rolleyes:

run some better tests ... 3dmark06 cpu score is not good for performance compare ...

run PCmark05, or some games ...

expreview ran 3D 06 because there were people requesting that bench

just like the crysis test,we have say that HT means nothing effi on real world test by destop system.

but anyone know why the HT is 1600MHz? on 200MHzX15 Phenom X4 system.

Quote:

---

Originally Posted by cookerjc

here comes 3Dmark06:

---

thank you for your contribution :up:

HT means something. it provides direct links to other pieces of hardware with out going through the FSB.

Quote:

---

Originally Posted by knightwolf654

HT means something. it provides direct links to other pieces of hardware with out going through the FSB.

---

thers no fsb on amd chips (at least not since k7...)

Quote:

---

Originally Posted by knightwolf654

HT means something. it provides direct links to other pieces of hardware with out going through the FSB.

---

Currently, on desktop setups, all of the bandwidth of HT1 is not fully utilized. HT3 is pointless on the desktop. The only arena where it provides a significant advantage is in 4P and up servers.

Quote:

---

Originally Posted by mstp2009

Currently, on desktop setups, all of the bandwidth of HT1 is not fully utilized. HT3 is pointless on the desktop. The only arena where it provides a significant advantage is in 4P and up servers.

---

strange
some members in here pointed out that HT1 become satured with only 2 cores

Quote:

---

Originally Posted by GoThr3k

strange
some members in here pointed out that HT1 become satured with only 2 cores

---

Proof please. Hearsay (especially from you) is worth nothing.

Quote:

---

Originally Posted by mstp2009

Currently, on desktop setups, all of the bandwidth of HT1 is not fully utilized. HT3 is pointless on the desktop. The only arena where it provides a significant advantage is in 4P and up servers.

---

give me proof.

Quote:

---

Originally Posted by mstp2009

Proof please. Hearsay (especially from you) is worth nothing.

---

it is posted a couple of times....
just use the search button...

EDIT// if you don't believe it,your problem...
dont complain because you are to lazy to search for it

Quote:

---

Originally Posted by GoThr3k

it is posted a couple of times....
just use the search button...

---

YOU made the claim. It is therefore YOUR responsibility to BACK IT UP.

I'm not your mamma little boy. I won't do your homework for you.

Quote:

---

Originally Posted by knightwolf654

give me proof.

---

Will link it when I return from work (I'm late right now and forums are blocked by the proxy at work).

Quote:

---

Originally Posted by mstp2009

Will link it when I return from work (I'm late right now and forums are blocked by the proxy at work).

---

and please dont show of a singlecore.....

just noticed. i dont know how much the 3.0GHz version will be, i am gunna gues around $400 but looking at those scores. who wants to pay $700 more for a QX9650 for only 1000 more points in 3dmark. intel is overpriced.

Quote:

---

Originally Posted by knightwolf654

just noticed. i dont know how much the 3.0GHz version will be, i am gunna gues around $400 but looking at those scores. who wants to pay $700 more for a QX9650 for only 1000 more points in 3dmark. intel is overpriced.

---

You think AMD is gonna sell a 3GHz Phenom for $400?! LOL

At this rate I doubt if they can get a 3GHz Phenom out before they reach 45nm...

Intel is overpriced? What about the $266 Q6600, which probably beats the 2.4GHz Phenom in most benchmarks.

intel is overpriced at the high end because there is no competition there

Quote:

---

Originally Posted by GoThr3k

intel is overpriced at the high end because there is no competition there

---

wtf are you talking about.... EE was always 999$ (1000€) since the introduction of the extreme editions with the P4... actually the lost money by keeping the price constant since 2004... (inflation)

the other 2 EE you see coming in 08 are more like über extreme editions and its only logical to settle them on a higher price spot.

Quote:

---

Originally Posted by Hornet331

wtf are you talking about.... EE was always 999$ (1000€) since the introduction of the extreme editions with the P4... actually the lost money by keeping the price constant since 2004... (inflation)

the other 2 EE you see coming in 08 are more like über extreme editions and its only logical to settle them on a higher price spot.

---

Why are you talking about EE?
EE is dead as far as i know

NB Freq should be much higher at these clocks (IT DOES MAKE A DIFFERENCE AS NB FREQ= L3 FREQ), these tests don't give us any new information, only that they could take an opteron and get the same results :down: :down:

we need results with proper NB speed, DDR3-1066 with the best stimings you can reach with stock chips rated for these speeds for K10 and some vorkfield scores using X38 with top DDR3 speeds supported officially by X38 mobos and ram sticks.

then we can draw some conclusions, but not with these worthles POS tests :shakes:

Quote:

---

Originally Posted by GoThr3k

Why are you talking about EE?
EE is dead as far as i know

---

uhn dont play dump...:down:
Core x Extreme = replacement for EE

save your flame baits for mstp2009

Quote:

---

Originally Posted by Hornet331

uhn dont play dump...:down:
Core x Extreme = replacement for EE

save your flame baits for mstp2009

---

i just lowered myself to your level.....
i was merely given an explanation why intel prices are considered high at the high end....

Quote:

---

Originally Posted by Hornet331

wtf are you talking about.... EE was always 999$ (1000€) since the introduction of the extreme editions with the P4... actually the lost money by keeping the price constant since 2004... (inflation)

the other 2 EE you see coming in 08 are more like über extreme editions and its only logical to settle them on a higher price spot.

---

Firstly Intel did not lose money on them, they just did not make as much. There is a huge difference.

The ability to keep the price at $999 all through the Intel price war gives an indication that they felt no need to be competitive with AMD in that segment, they did in all the others. That indicates that they were "overpriced" to us buyers or to put it from Intels point of view, had a large maRgin.

I don't think you can bring P4 EE's into it as Intel only really started the pricewar with the last P4's but mainly Core chips, we do not know whether they would have brought the price down or not for a weaker part.

Another way of looking at it is if you think they are not overpriced then you must think they are either underpriced or else the right price. I doubt you would think they are underpriced and the mere fact that they are at an artifical $999 that does not vary between models shows they are not at the correct price, so that only leaves one option, they are overpriced.

Regards
Andy

Quote:

---

Originally Posted by GoThr3k

i just lowered myself to your level.....
i was merely given an explanation why intel prices are considered high at the high end....

---

then your explanation is just plain wrong... why was intel chargeing 999$ for P4EE, PDEE etc. when amd "stomped" them.

If you where right they would charge 1500$ for a Extreme Core and not 999$

Quote:

---

Originally Posted by zakelwe

Firstly Intel did not lose money on them, they just did not make as much. There is a huge difference.

The ability to keep the price at $999 all through the Intel price war gives an indication that they felt no need to be compeitive with AMD in that segment, they did in all the others. That indicates that they were "overpriced" to us buyers or to put it from Intels point of view, had a large magin.

I don't think you can bring P4 EE's into it as Intel only really started the pricewar with the last P4's but mainly Core chips, we do not know whether they would have brought the price down or not for a weaker part.

Another way of looking at it is if you think they are not overpriced then you must think they are either underpriced or else the right price. I doubt you would think they are underpriced and the mere fact that they are at an artifical $999 that does not vary between models shows they are not at the correct price, so that only leaves one option, they are overpriced.

Regards
Andy

---

good point, but you seem to forget such products aren't priced like normal products, they are prestige products. They are not bound to any normal price fluctuations, like every other prestige product (rolex, etc)

hell the could charge even 1500$ or more, ppl would get them anyway cause for them money is no object. So i dont see there any "overpricing".

The only question I want answered. Who is this guy with the engineering sample so blatantly able to break his NDA?

Quote:

---

Originally Posted by Hornet331

then your explanation is just plain wrong... why was intel chargeing 999$ for P4EE, PDEE etc. when amd "stomped" them.

If you where right they would charge 1500$ for a Extreme Core and not 999$

---

i am talking about the present,i dont live in the past......

Quote:

---

Originally Posted by GoThr3k

i am talking about the present,i dont live in the past......

---

yes and now(in the present or better on the 12th of Nov.) they are charging 999$ for there top model like they allways did, i dont know what you want... (price gouging of e-tailers arn't intels fault)

Quote:

---

Originally Posted by User9498

The only question I want answered. Who is this guy with the engineering sample so blatantly able to break his NDA?

---

Who says he signed an NDA? ;)

Quote:

---

Originally Posted by Hornet331

yes and now(in the present or better on the 12th of Nov.) they are charging 999$ for there top model like they allways did, i dont know what you want... (price gouging of e-tailers arn't intels fault)

---

high end is not only the 999$ part....

and again,i was just explaining why intel's prices may considered high

Quote:

---

Originally Posted by GoThr3k

high end is not only the 999$ part....

and again,i was just explaining why intel's prices may considered high

---

show me one intel part that is more then 999$ till now.

the skulltrail cpus arnt gona be relead for a while and same goes for the 3.2ghz yorkfield, and as i mentioned erlier they are new products with a new price spot.

If they can charge more money for prestige products they will do it, i have no problem with that.

Also your labeling the category wrong, intel itself sees the extreme parts in the ethusiast category, which is setteld above the high end category...

and to repeat myself again, if you dont want to spend that much money on a cpu: DONT DO IT.

Noone forces you to buy it, there are plenty of cheap alternatives.

Quote:

---

Originally Posted by Hornet331

show me one intel part that is more then 999$ till now.

the skulltrail cpus arnt gona be relead for a while and same goes for the 3.2ghz yorkfield, and as i mentioned erlier they are new products with a new price spot.

If they can charge more money for prestige products they will do it, i have no problem with that.

Also your labeling the category wrong, intel itself sees the extreme parts in the ethusiast category, which is setteld above the high end category...

and to repeat myself again, if you dont want to spend that much money on a cpu: DONT DO IT.

Noone forces you to buy it, there are plenty of cheap alternatives.

---

i was talking about intel cpu's <999$
sigh,i am not going to repeat myself for the third time.....

and about the labbeling,i would call xe ultra enthusiast,the part below that enthusiast :rolleyes:
please,dont be ridicilous,you got my point (or so i hope:)

Please remember that, for now there is no native 3GHz Phenom. The test was OC the 2GHz ES(HT 1600MHz).

Honestly, it was a little closer than expected. As far as the ram speed, he seemed to manage to increase it for the Phenom in the Crysis test. Why not run the same fsb and timmings with 3dmark. I don't understand why he didn't for this test, but maybe more comming later.

Quote:

---

Originally Posted by expreview

Please remember that, for now there is no native 3GHz Phenom. The test was OC the 2GHz ES(HT 1600MHz).

---

and there is no way to overclock the northbridge (L3 cache)?

2 ghz opteron also has 1800 mhz NB speed using split power planes

Quote:

---

Originally Posted by generics_user

and there is no way to overclock the northbridge (L3 cache)?

2 ghz opteron also has 1800 mhz NB speed using split power planes

---

BIOS have no option to change NB FID.The ES is 8x,so if the 200MHz FSB,NB Core speed only have 1600MHz Option.

some guy does not want to see the ES which can OC 3GHz, so let me do another test with the 9600/9500 retail.....

i love phenom, but these split power lanes, weird ass dividers, crap, I might go Intel just so I know what everything is doing!

God, your an annyoing and inconsequent person....

first you say that:

Quote:

---

Originally Posted by GoThr3k

intel is overpriced at the high end because there is no competition there

---

this implayes that you are complaining about the extreme cores (cause a Q6600 for 211€ or a Q6700 for 450€ and a E6850 for 230€ isnt overpriced.)

than you suddenly say

Quote:

---

Originally Posted by GoThr3k

i was talking about intel cpu's <999$
sigh,i am not going to repeat myself for the third time.....

---

so if your not talking about the enthusiast cpus, which of the cpus are overpriced, tell me? the Q6700?

Quote:

---

Originally Posted by GoThr3k

and about the labbeling,i would call xe ultra enthusiast,the part below that enthusiast :rolleyes:
please,dont be ridicilous,you got my point (or so i hope:)

---

say that to intel, extreme cores are labled as enthusiast cpus, period.

Quote:

---

Originally Posted by GoThr3k

i just lowered myself to your level.....
i was merely given an explanation why intel prices are considered high at the high end....

---

oh the name calling... :rolleyes:

allways the same people.

the first post you quoted was a reply to the post above that....
someone else complained about the high prices,i just gave an plausible explanation.....

and btw, thats not name calling,that was reality.....

you cant stop... its really sad...

Why are the Intel rigs getting a higher SM score? I thought that was all GPU, guess not.

OMG, the :banana::banana::banana::banana::banana:ing is just getting ridiculous..

Quote:

---

Originally Posted by Hornet331

uhn dont play dump...:down:
Core x Extreme = replacement for EE

save your flame baits for mstp2009

---

If you drink water with your breakfast, then replace it by milk, water won't become milk, but you still have a drink.

In this case, high-end stays, the EE doesn't.

well, its gonna be really cheap im guessing if they plan on selling alot. it looks like they are targeting people who dont wanna spend the $266 for a intel cpu.

Quote:

---

Originally Posted by Calmatory

If you drink water with your breakfast, then replace it by milk, water won't become milk, but you still have a drink.

In this case, high-end stays, the EE doesn't.

---

point taken, but how can you say its expensive?

lets say E4400 is water and E6850 is milk? if you call 210€ for the top of the line dualcore expensive i think your not involved that long with cpus. ;)

Quote:

---

Originally Posted by knightwolf654

give me proof.

---

You wanted proof, here it is:
http://www.hypertransport.org/tech/index.cfm
http://www.hypertransport.org/images/features_chart.gif


HT1 has a max BW of 12.8GB/s at 800MHz operation.

Considering that we are talking ONE SOCKET SYSTEMS (Phenom, not Barcelona), that means even AFTER you subtract out the 6.4GB/s for full utilization of a dual-channel DDR2 controller on the memory bus, there is a full 6.4GB/s left over.

All CORE-TO-CORE COMMUNICATION on X2 and Phenom CPUs occurs via the crossbar or L3 cache, respectively. So there is STILL nothing that can fully utilize the extra 6.4GB/s BW in HT1, since CORE-TO-CORE COMMUNICATION has a dedicated pathway.


EDIT - HT3 on the desktop is like going from a 4 lane to a 16 lane highway when you are the ONLY car on the road in both cases. Just because you have more lanes, it doesn't magically make your car faster when there is nothing to impede it in the first place.

Quote:

---

Originally Posted by mstp2009

You wanted proof, here it is:
http://www.hypertransport.org/tech/index.cfm
http://www.hypertransport.org/images/features_chart.gif


HT1 has a max BW of 12.8GB/s at 800MHz operation.

Considering that we are talking ONE SOCKET SYSTEMS (Phenom, not Barcelona), that means even AFTER you subtract out the 6.4GB/s for full utilization of a dual-channel DDR2 controller on the memory bus, there is a full 6.4GB/s left over.

All CORE-TO-CORE COMMUNICATION on X2 and Phenom CPUs occurs via the crossbar or L3 cache, respectively. So there is STILL nothing that can fully utilize the extra 6.4GB/s BW in HT1, since CORE-TO-CORE COMMUNICATION has a dedicated pathway.


EDIT - HT3 on the desktop is like going from a 4 lane to a 16 lane highway when you are the ONLY car on the road in both cases. Just because you have more lanes, it doesn't magically make your car faster when there is nothing to impede it in the first place.

---

:rofl: awesome proof you got there :rofl:

Quote:

---

Originally Posted by GoThr3k

:rofl: awesome proof you got there :rofl:

---

Reading comprehension is a B!tch isn't it?

Quote:

---

Originally Posted by mstp2009

Reading comprehension is a B!tch isn't it?

---

http://www.xbitlabs.com/images/cpu/a...0/platform.jpg
http://www.septor.net/specs/amd/img/AMD8151BlockL.jpg
http://www.hothardware.com/articleim.../blockdiag.png

especially when memory doesnt have anything to do with hypertransport.....

doesnt look like the memory controller is connected with a hypertransport bus

Quote:

---

Originally Posted by GoThr3k

http://www.xbitlabs.com/images/cpu/a...0/platform.jpg
http://www.septor.net/specs/amd/img/AMD8151BlockL.jpg
http://www.hothardware.com/articleim.../blockdiag.png

especially when memory doesnt have anything to do with hypertransport.....

doesnt look like the memory controller is connected with a hypertransport bus

---

GREAT!

That just makes my argument THAT MUCH STRONGER.


Core to core communication doesn't use HT either on 1 socket systems. So THERE GOES THAT THEORY.
If the memory bus isn't sharing HT bandwidth, then that what ON EARTH do you possibly expect to consume >12.8GB/s of bandwidth?

Quote:

---

Originally Posted by mstp2009

GREAT!

That just makes my argument THAT MUCH STRONGER.


Core to core communication doesn't use HT either on 1 socket systems. So THERE GOES THAT THEORY.
If the memory bus isn't sharing HT bandwidth, then that what ON EARTH do you possibly expect to consume >12.8GB/s of bandwidth?

---

haha,first you are providing proof
proof isnt correct at all,know you say it confirms that your proof is right
:rofl: :rofl:

Are there genuine gripes here or are people just mad AMD is getting smoked? :welcome:

Quote:

---

Originally Posted by GoThr3k

haha,first you are providing proof
proof isnt correct at all,know you say it confirms that your proof is right
:rofl: :rofl:

---

Still waiting on YOUR proof that HT2/3 is needed to overcome some mythological bottleneck.


I'll HAPPILY admit that having the memory bus on the HT bus was an error on MY PART. But still doesn't INVALIDATE the point that on a SINGLE SOCKET SYSTEM, there is bandwidth to spare on the HT bus.

That is the SINGLE POINT you have trouble grasping (that poor reading comprehension thing again - bites you in the arse every time).

Quote:

---

Originally Posted by Kingcarcas

Are there genuine gripes here or are people just mad AMD is getting smoked? :welcome:

---

I'm pissed, for one, that K10 only comes up to par with an architecture that has been out 16 months now. It means we are all going to get d!cked over at the cash register when we buy high-end equipment.

GoThr3k just has this unfounded idea that HT3 is magically going to make K10 a faster CPU. HT3 doesn't do SQUAT unless you are in a multi-socket system.

Quote:

---

Originally Posted by mstp2009

I'm pissed, for one, that K10 only comes up to par with an architecture that has been out 16 months now. It means we are all going to get d!cked over at the cash register when we buy high-end equipment.

GoThr3k just has this unfounded idea that HT3 is magically going to make K10 a faster CPU. HT3 doesn't do SQUAT unless you are in a multi-socket system.

---

QFT

however NB speed in K10 has a HUGE impact on performance and if i understand it correctly ht link speed + NB speed is the same on HT3 the HT link speed does have a significant impact on performance with K10 (not due to limited I/O bandwidth more due to a higher NB+L3 frequency -> more effective prefetchers and lower latency L3 cache

Quote:

---

Originally Posted by generics_user

QFT

however NB speed in K10 has a HUGE impact on performance and if i understand it correctly ht link speed + NB speed is the same on HT3 the HT link speed does have a significant impact on performance with K10 (not due to limited I/O bandwidth more due to a higher NB+L3 frequency -> more effective prefetchers and lower latency L3 cache

---

Interesting on the lower latency. Can you link me where you read that?

Quote:

---

Split power planes. Feeding the memory controller and the core from different power rails is not a direct improvement to the memory subsystem, but it does allow the memory controller to be clocked higher than the CPU core.

---

Link
So modifying the NB speed is more an energy efficiency thing than an OC feature.;)

Quote:

---

Originally Posted by mstp2009

Interesting on the lower latency. Can you link me where you read that?

---

Kyosens latency (1800mhz)
some more posts ago this user said that his IMC runs at 1400 mhz-note the higher latency

and as we know K10 IMC speed = NB speed

Quote:

---

Originally Posted by justapost

Link
So modifying the NB speed is more an energy efficiency thing than an OC feature.;)

---

not really, NB SPEED = L3 speed...

Quote:

---

Originally Posted by expreview

BIOS have no option to change NB FID.The ES is 8x,so if the 200MHz FSB,NB Core speed only have 1600MHz Option.

some guy does not want to see the ES which can OC 3GHz, so let me do another test with the 9600/9500 retail.....

---

I'm sorry for you that you provide some long waiting results and that some people just complain they are not like they have dream. :down:
Thank you anyway for your work ;)

Quote:

---

Originally Posted by User9498

The only question I want answered. Who is this guy with the engineering sample so blatantly able to break his NDA?

---

Why, you'd prefer to have only inquirer informations?

NB communicates to the CPU via HT, it's been this way since HT1.

But the L3 talks to memory, not the NB.

Sure you aren't looking at memory latency? Those latencies are horrible for a cache (even L3).

EDIT - I was referring to the first link which talks about ns time, fyi. The second one clearly shows cycles (which at 52 cycles is not pretty also).

mstp2009 lets wait for the 19th/20th i think then there is plenty of information available to discuss stuff like this. :)

Quote:

---

Originally Posted by mstp2009

NB communicates to the CPU via HT, it's been this way since HT1.

But the L3 talks to memory, not the NB.

Sure you aren't looking at memory latency? Those latencies are horrible for a cache (even L3).

EDIT - I was referring to the first link which talks about ns time, fyi. The second one clearly shows cycles (which at 52 cycles is not pretty also).

---

kyosen got 43 cycles with 1800 mhz on the L3 cache:

kyosen's screenie

still not pretty but it shows what i mean ;)

Quote:

---

Originally Posted by expreview

BIOS have no option to change NB FID.The ES is 8x,so if the 200MHz FSB,NB Core speed only have 1600MHz Option.

---

Ah, ok it was not mentioned in the article what stock clock the phenom has. Thank you for the clarification.

Quote:

---

Originally Posted by generics_user

not really, NB SPEED = L3 speed...

---

Ok, from an extreme perspective that has an impact. :rolleyes:
I thought about a more efficient memory sytem at lower cpu clocks.

Quote:

---

Originally Posted by mstp2009

I'm pissed, for one, that K10 only comes up to par with an architecture that has been out 16 months now. It means we are all going to get d!cked over at the cash register when we buy high-end equipment.

GoThr3k just has this unfounded idea that HT3 is magically going to make K10 a faster CPU. HT3 doesn't do SQUAT unless you are in a multi-socket system.

---

HT3 will do nothing much for K10 (in single socket systems) i have never stated otherwise
maybe 1-2&#37;,but that me be recognised as statistical error ;)

Quote:

---

Originally Posted by GoThr3k

HT3 will do nothing much for K10 (in single socket systems) i have never stated otherwise
maybe 1-2%,but that me be recognised as statistical error ;)

---

AGREED.

The cycle change in latency b/w the two users listed above is VERY interesting. I didn't check, but were they the same stepping of K10?

43 cycles, however, is still play ugly.

Quote:

---

Originally Posted by GoThr3k

HT3 will do nothing much for K10 (in single socket systems) i have never stated otherwise
maybe 1-2%,but that me be recognised as statistical error ;)

---

yes HT 3.0 is not so usefull for single socket, but don't forget, HT 3.0 motherboards come with split plane, and increase the memory controler frequency. ;)

Quote:

---

Originally Posted by mstp2009

AGREED.

The cycle change in latency b/w the two users listed above is VERY interesting. I didn't check, but were they the same stepping of K10?

43 cycles, however, is still play ugly.

---

It's a new serial cache. It's a bit different from old cache but it's as fast even the high cycles latency.

Brisbane use the same cache for L2 ( K10 for L3 ). It's high latency but it's as fast. :yepp:

Quote:

---

Originally Posted by mstp2009

AGREED.

The cycle change in latency b/w the two users listed above is VERY interesting. I didn't check, but were they the same stepping of K10?

43 cycles, however, is still play ugly.

---

both B1 check cpu-z screens.

as L3 cache seems to handle cache traffic between the cores (L2) and serves as a buffer between the CPU and the IMCs there should be some more % achieveable by increasing NB freq...

and we still don't know if the IMC runs in ganged or unganged mode (unganged allows the IMC to write and read at the same time opposed to read/write only in old K8 IMC, at least that's how i understand it after reading kyosens thread), if the IMC runs ganged there is a chance that K10 can get a higher IPC than yorkfield (but i don't count on it :rolleyes: )

Quote:

---

Originally Posted by generics_user

both B1 check cpu-z screens.

as L3 cache seems to handle cache traffic between the cores (L2) and serves as a buffer between the CPU and the IMCs there should be some more % achieveable by increasing NB freq...

and we still don't know if the IMC runs in ganged or unganged mode (unganged allows the IMC to write and read at the same time opposed to read/write only in old K8 IMC, at least that's how i understand it after reading kyosens thread), if the IMC runs ganged there is a chance that K10 can get a higher IPC than yorkfield (but i don't count on it :rolleyes: )

---

IIRC doesn't the L3 cache and IMC scale their MHz/GHz speed differently depending on the CPU multiplier and the relative frequencies of the cores?

There was some complex BS going on w/ K10 that made it pretty decent (on par w/ C2Q) at certain frequencies, and absolutely SUX at others.

I will show you guy the Retail version test with GP9600, like Cinebench score will announce soon.

you would know about the Phenom default performance.....

Quote:

---

Originally Posted by expreview

I will show you guy the Retail version test with GP9600, like Cinebench score will announce soon.

you would know about the Phenom default performance.....

---

the new CPU-Z Version can show NB speed and correct memory speed and timing...

here...version CPU-Z Version 1.41.4 preview

http://img232.imageshack.us/img232/3...0400pr8.th.png