Where is AMD?!?

[stealth17]

WTF is up with AMD? Are they dead?

Here is is Jan 9, 2007. I've only seen one brisbane, no K8L. They aren't doing :banana::banana::banana::banana: lately :upset:

[freeloader]

Well, their timetable doesn't revolve around us so I'm pretty sure they just don't give a rat's ass! They'll continue to lose many enthusiasts to Intel if they don't get their act together.

[Omastar]

AMD execs are too busy having sex with former ATI execs. Processors will come later!

"Unf, oh God, I'm merrrgiiiing!"

[gullf1sk]

Maybe they are not confident in their product.
When Intel had Conroe prototypes ready, they couldnt wait to let people see what they had.
I dont have much faith in K8L.. or should i say, KL8

[Amey]

^^ Isnt it too early to jump to conclusions ? Remember, Intel Launched C2D when they wanted not when we wanted. So isnt that applicable to AMD too ?

[cky2k6]

Maybe they are not confident in their product.
When Intel had Conroe prototypes ready, they couldnt wait to let people see what they had.
I dont have much faith in K8L.. or should i say, KL8
first off, your avatar is great... but yeah, i dont know about k8l, it might flop, but have you not seen rd790 specs? same thing with quad fx vs v8 really, intel doesnt have any friends in the multi-gpu space... chances are, even though intel may have the superior cpu, which is again debatable since we havent seen k8l which is not looking awfully promising, but anyways intel needs to partner up with nvidia fast or amd/ati will recover alot more of the high end market then intel will anticipate with quad fx on a reasonable chipset.

[Serge84]

If K8 was a quad, it would perform like QFX. Its not that far behind the C2Q btw. Thats why AMD doesn't need a better quad yet, why would it when the 2 perform almost the same? AMD's direct connection arc made quad core and octo core before intel ever did anything great with 2 sockets. Whats the difference with glueing 2 peices of die together, on one socket, compared to 2 dual cores on 2 sockets? It makes no sense to say what happened to AMD when they work in the darkness unlike Intel always showing what they can do.

Perfect for AMD to work under cloak and dagger eh! AMD is 2nd best, so what even by a little performance difference. In the real world you'll never tell. So theres no point in saying OMG, amd lost by 5-15%... what the f am I going to do my life is over... Oh wait spend $5000 on a Intel system and get to the performance I been missing. >_> yeahhh riggght... silly point to blow off a lot of money to be a show off is all your doing. AMD is doing just fine and always has been just fine 2nd best before K8 was around. Just like Intel was just fine for 3 years being 2nd best until conroe. Companies that are 30+ years old don't die over night or even lose anything at all. Both are going to be around for a long time. AMD is not going to drop off the map just because a cpu out performs its cpu for a little wile for 6 months or a year even. Won't make any difference at all, they still make amazing sales still being 2nd best. Get over it god. Going to croak or something just because AMD's best is behind only a little? So childish...

Seems to make no difference at all what they are or how they are connected, they are both linked to eachother in a similar way. Intel uses one board to put 2 dies together on one socket limited by a FSB, but AMD uses 2 boards on 2 sockets with no limit because of the HTL bus, so because they are divided by sockets doesn't mean squat. None are native quads or true quad cores. *Shrugs* So whats the point? Everything is software limited. So think how K8L will perform over C2Q. But AMD is not the type to brag, they wait and seal there lips shut until they release the chip. Its not LATE. Its sipose to come out in the 2nd half of 2007. AMD never said it was ever sipose to come out now. Get your facts together.

[Brother Esau]

Don't get you're Knickers in a twist dude!
My money is on the unified AMD/ATI and thats what is taking so long but when it does hit Conro will just be code name for some guy in cell block D:stick:

[squilliam]

You fan boys are annoying. Do you whack of to this every night?
I'd recommend something other than caffeine for you.
Someone please enlighten me on how two x2 3800s are cheaper
than a E6400. I'd really like to know. Oh and I am pretty sure
dual socket mobos aren't that cheap either. Can you get SLI
with a Dual socket mobo? That's right, no you can't.

BTW The opteron 165 isn't built the same as the FX-62.
It handles differently. So quit the fantasy's and quit blowin
smoke up each other's arses.

[KingDingeling]

agreed with squilliam, this horse has been beaten to death more than a thousand times on thousands of forums including XS. AMD will have something coming, but they have bigger fish to fry at the moment, such as the fusion with ATI, that definetly takes more than 1 day to organize etc.

[Brother Esau]

agreed with squilliam, this horse has been beaten to death more than a thousand times on thousands of forums including XS. AMD will have something coming, but they have bigger fish to fry at the moment, such as the fusion with ATI, that definetly takes more than 1 day to organize etc.

Yea..That:woot:

[Equil|briuM]

You fan boys are annoying. Do you whack of to this every night?
I'd recommend something other than caffeine for you.
Someone please enlighten me on how two x2 3800s are cheaper
than a E6400. I'd really like to know. Oh and I am pretty sure
dual socket mobos aren't that cheap either. Can you get SLI
with a Dual socket mobo? That's right, no you can't.

BTW The opteron 165 isn't built the same as the FX-62.
It handles differently. So quit the fantasy's and quit blowin
smoke up each other's arses.

It's an AMD board, and your telling people to stop going on about it? :rolleyes:

Dual Socket Mobos and SLi have been out for AGES.

3800+ X2's are coming down in price all the time - you can pick them up for around £80 when a E6400 is £150.

So whats your answer?

BUY A CONROE!

Yeah, whatever - let me know when the extra £80 odd is worth the -10s In PI.

Granted, if your gunna purchase a new system and are WILLING to spend around that amount, Conroe is the choice, only a fool would think different.

[Serge84]

There is a limit you know where benchmarks don't tell the whole story like in the real world. It doesn't matter how fast you make your rig, its going to be so fast you can't tell the difference. Software is one thing thats limiting stuff. 2nd of all is once you get to a fast enough speed everything opens as quick as you can notice them as soon as it gets clicked it opens already loaded. In that

Area AMD and Intel are matched far beyond what software can keep up with now adays. For both anything over 3ghz is not noticable visually in a game regardless how much your benchmark score is. Its only to show off and useless to you when visually the software or game is just maxed out where you can't see the difference. XF's have been like that for a wile. Same with XE's both are great chips and don't improve anything better in the real world that you can notice is what I'm trying to point out.

Great, good for you that your cpu can get 0.9 sec on S-PI. Can't tell the difference from a 3800+ X2 OCed to 3ghz. Conroe in the real world is being held back by FSB and software. AMD software limited as well. Or as simple as visually limited. No matter how you play your cards with a benchmark are you going to beable to tell the difference between 2 rigs unnamed? No!

Thats my drift, get it... both no matter what you pick are going to blow your socks off and feel no different in the real world to a point. About 3ghz for both is useless for anything faster then that. And Quad cores are useless at the moment but in the future only time will tell. We don't need a vs contest in every forum. Both will feel as good as it gets. AMD and Intel are both neck on neck, and hardly worth going mad over for a few % or going to make any difference at all in the real world. Wait until they are 50% appart or 100% appart and software uses 100% of dual cores then maybe we will start seeing a much bigger difference in the future of the real world but for now its questionable.

What you need is a GPU upgrade if your limited in games, your cpu most of the time is hardly even stressed. Your video card does all the work in games not your CPU.

[KingDingeling]

Serge84, I will be upgrading to Conroe soon because in things like DVD De/Encoding faster is better and whilst editing videos/photochopping faster CPU and fast RAM is best!

[Serge84]

Thats great for you because you want to save a little time, but some of us don't really care about if it takes a extra 30 secs or 30 mins. Getting stuff done 5 to 15% faster just isn't worth the money to others. I'd like the small performance jump sure but time is not a critical thing in my world. Is it going to make that much of a difference? Is it really worth the extra money for your job to take that little amount of time off a encoding or something?

What your real limit is, is you need more... faster memory, faster hard drives, and a faster dvd burner. That makes the difference more then a faster cpu. Because you forget data is only as fast as it can be read/writen to the hard drive... most of all that is everybodys hardware limit. Thats how people really get faster encodings, with raptors in raid.

[gallag]

Thats great for you because you want to save a little time, but some of us don't really care about if it takes a extra 30 secs or 30 mins. Getting stuff done 5 to 15% faster just isn't worth the money to others. I'd like the small performance jump sure but time is not a critical thing in my world. Is it going to make that much of a difference? Is it really worth the extra money for your job to take that little amount of time off a encoding or something?

What your real limit is, is you need more... faster memory, faster hard drives, and a faster dvd burner. That makes the difference more then a faster cpu. Because you forget data is only as fast as it can be read/writen to the hard drive... most of all that is everybodys hardware limit. Thats how people really get faster encodings, with raptors in raid.

You have to admit though; it’s pretty funny that since Conroe came out performance seems to have stopped mattering in the amd camp lol.

[Omastar]

I'm surprised someone hasn't brought up the fact that AMD is still faster in Science Mark. Yes, thank God we have Science Mark! Threads like this make my heart hurt.

[alucasa]

I'm surprised someone hasn't brought up the fact that AMD is still faster in Science Mark. Yes, thank God we have Science Mark!

hehe, may I mention that amd also catches up in 64bit.

Intel and AMD are both good for me. For so, I use cpus from both companies.

[nn_step]

1) brisbane is very popular right now with the OEMs (so you'll have to wait until they have excess stock)
2) K8L isn't due until 2H 2007. Please be patient. It would be kinda like me back January 2006 and :banana::banana::banana::banana::banana:ing about how conroe isn't out yet

[kiwi]

There are things that Intel (or maybe software?) needs to optimize, one is their 64bit extension EMT64. Opteron @ 3ghz beat conroe 3.6ghz in 64bit openssl key generation on linux :) Sure it is more server side application but still, I was very surprised myself.

[arisythila]

I play games... and for games the extra 5 FPS isnt worth the cost of a new Evga 680i, and a conroe.

~Mike

[[XC] Teroedni]

There are things that Intel (or maybe software?) needs to optimize, one is their 64bit extension EMT64. Opteron @ 3ghz beat conroe 3.6ghz in 64bit openssl key generation on linux :) Sure it is more server side application but still, I was very surprised myself.

I very much like to see some proof for that
Links Please:D

[n91htmare]

You have to admit though; it’s pretty funny that since Conroe came out performance seems to have stopped mattering in the amd camp lol.

haha yah, tell me about it. All of sudden the Benchmarks are rigged. Benchers are paid by Intel. Benchmark guidelines should change.
We don't care about performance any more like you said. oh we don't care about 5% speed difference but when A64 was 2% faster, it was fraggin intel.
How fast the tables turn.

K8 never had as much performance over P4 as Conroe has over k8.

I look at old benchmarks, Intel was still VERY CLOSE or ahead in media benchmarks. Gaming is what killed it.

I look at conroe and k8, well, different ballgame.

It's ok to accept. it's not gonna belittle you.

I jump ships to whatever gives me the best bang for my buck. I don't care it's Intel, AMD, or whoever.
I had a p4 3.0..... i saw how the opty165 was kickin arse and overclocking insanely. I got myself an opty 165.
I then saw the core 2 duo, liked the encoding advantage of it, had the $$$ and bought it.
If i see that k8L is the next coming, if I have $$ i'll get k8l and praise AMD....

If intel sucked 2 years ago, i have no problem saying it blows. i don't try to defend it because i have NO A THING TO GAIN by defending it. If anything, i'll be paying more $$ for a lower performing product.

[kemo]

I play games... and for games the extra 5 FPS isnt worth the cost of a new Evga 680i, and a conroe.

~Mike
Off topic
Why 450$ for the memory?

[n91htmare]

Off topic
Why 450$ for the memory?

Because apparently, spending $450 on memory to be used on an AMD system is more justifiable somehow?

And the Performance Boost he got from the $450 memory is also some how worth the cost.

[kiwi]

I very much like to see some proof for that
Links Please:D

Those are tests done by myself :)

But to give you an idea here are some numbers that are left:

opteron 3Ghz:

sign verify sign/s verify/s
rsa 512 bits 0.0002s 0.0000s 5892.1 67618.3
rsa 1024 bits 0.0005s 0.0000s 1857.7 30162.6
rsa 2048 bits 0.0028s 0.0001s 354.0 11255.8
rsa 4096 bits 0.0173s 0.0003s 57.7 3560.6

Conroe @ 3.7ghz

sign verify sign/s verify/s
rsa 512 bits 0.000164s 0.000012s 6093.1 81953.4
rsa 1024 bits 0.000576s 0.000032s 1736.1 31463.7
rsa 2048 bits 0.003274s 0.000094s 305.4 10648.0
rsa 4096 bits 0.020682s 0.000313s 48.4 3191.4

Look at the 1st and the 2nd column from right, more is better!

One explanation might be newer openssl version but results cannot be so much different, a few 4096 keys/s at most. Another is that AMD 64bit implementation is better or software is optimized for that. Bear in mind 64bit version is way better than 32bit :)

You can try running yourself openssl if you have it:
openssl speed rsa dsa

[Shintai]

Those are tests done by myself :)

But to give you an idea here are some numbers that are left:

opteron 3Ghz:

sign verify sign/s verify/s
rsa 512 bits 0.0002s 0.0000s 5892.1 67618.3
rsa 1024 bits 0.0005s 0.0000s 1857.7 30162.6
rsa 2048 bits 0.0028s 0.0001s 354.0 11255.8
rsa 4096 bits 0.0173s 0.0003s 57.7 3560.6

Conroe @ 3.7ghz

sign verify sign/s verify/s
rsa 512 bits 0.000164s 0.000012s 6093.1 81953.4
rsa 1024 bits 0.000576s 0.000032s 1736.1 31463.7
rsa 2048 bits 0.003274s 0.000094s 305.4 10648.0
rsa 4096 bits 0.020682s 0.000313s 48.4 3191.4

Look at the 1st and the 2nd column from right, more is better!

One explanation might be newer openssl version but results cannot be so much different, a few 4096 keys/s at most. Another is that AMD 64bit implementation is better or software is optimized for that. Bear in mind 64bit version is way better than 32bit :)

You can try running yourself openssl if you have it:
openssl speed rsa dsa

Care to elaborate on version and OS used, along with details of the actual systems used? I´m just curious.

[Lenny_Nero]

I have just built a very quick (IMO) AMD x2 box for dirt cheap money, I also got some bits from someone that had done the upgrade/new build to conroe (from an nF4 and opty at 2.9+ GHz) and he said 'it was not that much faster, if I had known it was only going to be this much I might not have done it'. I also have a few friends of the family that have not long got 'of the shelf' conroe boxes and have both said that they expected a lot more.

What I also find funny is the way the conroe users/want'ers are so quick to call anyone who says anything good about AMD 'fanboys', maybe I need to understand the definition better :)

As for what Serge84 said, I agree with 100%.

I also know of many people that are very happy with their Opterons and x2's that do faster than FX62 speeds and none (as far as I have seen) have ever been under any illusions that they have an FX62 or a chip built the same way, just (as said) one that goes as fast (or faster) for a lot less money. I have got a 4200x2, G.Skill and DFI board for less than ~£250, you cant do that with conroe,

[SunFlowerSeeds]

That is precisely what I am feeling. X2 3800+ paired with X1900XT running BF2148 at 1280x1024, 6AA, 16AF with ATMAlphaSharpenmode + EAMT turn on without any sweat...


I couldn't have said it better myself. AMD will go on regardless of what conroe is doing now. AMD dual cores are fast enough for our daily needs at a amazing price, heck I only spent 150 dollars for my opty165 and now its running at FX62 speeds and chews up games like nothing.[QUOTE]

That going to make lots of people angry... :p:

[QUOTE=Mike6969]Why throw away or switch to something like 15% faster? Waste money and time when you would never notice a such improvement , unless you're very bored at home and got nothing much to do. :woot:

[arisythila]

Off topic
Why 450$ for the memory?

Mainly for my e-penis. (Just joking) I wanted to see how hard it would be to reach 3GHz with my processor on air. With decent memory.

Im actually having some problems due to my video card.. (PIECE OF :banana::banana::banana::banana: 8800GTX) So EVGA has one comming to me now, After that, I will test what I can do.

~Mike

[arisythila]

Because apparently, spending $450 on memory to be used on an AMD system is more justifiable somehow?

And the Performance Boost he got from the $450 memory is also some how worth the cost.

My OCZ PC2-8000 fried @ 2.8v DDR2-1200 4-4-4-15 So I had to buy some more memory. This was in my price range.

~Mike

[awdrifter]

We enthusiasts only represents a very small percent of AMD's sales, and for the majority, they don't oc. So AMD's cpus are still very competitive (compared to the similarly priced Conroe at stock clock, cuz for the same price of a E6300, you can get a X2 4200). So AMD is in no hurry to bring out K8L.

[SnipeR²]

Honestly, i own an AMD based rig. But if i had to build a system from scratch, i would go with a conroe/C2D (as i did for a client) because right now they are cheap and perform really well. However, i wouldnt upgrade to a conroe from my current rig just to get a higher figure in a benchmark. That's just unnecessary expenses and my rig already eats anything that is thrown at it. There are many AMD rigs that are top end, and do not need upgrades...

[gallag]

My OCZ PC2-8000 fried @ 2.8v DDR2-1200 4-4-4-15 So I had to buy some more memory. This was in my price range.

~Mike


jeeeze, for that kind of money you probably could of done a Conroe upgrade and got a good 40% faster performance ( overclocked of course) :D

But seriously it is amazing how You could be spinning this story of how core 2 is not a viable upgrade for 15-20% difference (at stock) yet is seems a worthwhile investment to spend more on ram to get an extra 10% out of a amd lol

Surely you can see what we are getting at here, why do you just not say "sure the Intel is 15-40% faster (both fully maxed out) but I prefer amd so i am going to stick with them" instead of telling people that that they would not notice a difference between the two, like there have been loads of people that have upgraded from a low end x2 to a high end x2 and noticed a difference it is how it has always been but all of a sudden performance has stopped mattering (at least till k8l comes)

I mean last year these forums were full of posts like “I changed my 4200 to a 4800 and got an extra 200 MHz whoohoo" and not one person on extreme systems said "what a waste"

Also I really hope that k8l murders core 2 simply because I can then buy one and smite any Intel fanboy that dares say that is worthless because it is only 10-25% faster and prove that I am not a fanboy except of performance

Anyway you probably will have some clever reply and that is fine, I know I will not change your mind nor do I want to so this is the last I will reply in this thread but just try to open your mind and look at it without the green glasses on.

Regards gallag

[kiwi]

Care to elaborate on version and OS used, along with details of the actual systems used? I´m just curious.

opteron 146 @ 3ghz, mushkin redline 2-2-2-5 250Mhz, DFI Expert
OS: Ubuntu 64bit linux (latest version available, march 2006)

C2D E6600 @ 3.7ghz, p5b dlx, cellshock 2x1GB @ 463 4-4-4
OS: FC6 64bit

[nn_step]

opteron 146 @ 3ghz, mushkin redline 2-2-2-5 250Mhz, DFI Expert
OS: Ubuntu 64bit linux (latest version available, march 2006)

C2D E6600 @ 3.7ghz, p5b dlx, cellshock 2x1GB @ 463 4-4-4
OS: FC6 64bit
now that is a problem.
You really should have used matching Operating systems. Otherwise the result can't really be considered valid.

[arisythila]

jeeeze, for that kind of money you probably could of done a Conroe upgrade and got a good 40% faster performance ( overclocked of course) :D

I bought it because first K8L's are going to be DDR2, Probably with DDR2-1000 the standard. So it will be nice to have memory that can do DDR2-1300. for Overclocking purposes.



But seriously it is amazing how You could be spinning this story of how core 2 is not a viable upgrade for 15-20% difference (at stock) yet is seems a worthwhile investment to spend more on ram to get an extra 10% out of a amd lol


I don't "invest" into Intel for personal reasons. Probably the same reason I'm in law suit with them for 1.7 million dollars. It does look like I am on the winning side tho.

~Mike

[arisythila]

now that is a problem.
You really should have used matching Operating systems. Otherwise the result can't really be considered valid.

Neh, depends on what kernel he was using. They should be the same.

~Mike

[squilliam]

There are things that Intel (or maybe software?) needs to optimize, one is their 64bit extension EMT64. Opteron @ 3ghz beat conroe 3.6ghz in 64bit openssl key generation on linux :) Sure it is more server side application but still, I was very surprised myself.

Correct me if I am wrong, but isn't the opteron a server chip?
A standard Conroe however, is not.

[kiwi]

Opteron is server but so is Xeon. And xeon s775 does not beat regular conroe at the same speed since chips are identical except lower default vcore for xeon.


now that is a problem.
You really should have used matching Operating systems. Otherwise the result can't really be considered valid.

I have tested a lot of openssl and linux versions :) As I mentioned before the difference is a few 4096 keys/s at most

Anyway I have venice available and I can retest it on the same OS and openssl version but I am pretty confident venice will win this test, at least at the same clock speeds :)

[Serge84]

Those are tests done by myself :)

But to give you an idea here are some numbers that are left:

opteron 3Ghz:

sign verify sign/s verify/s
rsa 512 bits 0.0002s 0.0000s 5892.1 67618.3
rsa 1024 bits 0.0005s 0.0000s 1857.7 30162.6
rsa 2048 bits 0.0028s 0.0001s 354.0 11255.8
rsa 4096 bits 0.0173s 0.0003s 57.7 3560.6

Conroe @ 3.7ghz

sign verify sign/s verify/s
rsa 512 bits 0.000164s 0.000012s 6093.1 81953.4
rsa 1024 bits 0.000576s 0.000032s 1736.1 31463.7
rsa 2048 bits 0.003274s 0.000094s 305.4 10648.0
rsa 4096 bits 0.020682s 0.000313s 48.4 3191.4

Look at the 1st and the 2nd column from right, more is better!

One explanation might be newer openssl version but results cannot be so much different, a few 4096 keys/s at most. Another is that AMD 64bit implementation is better or software is optimized for that. Bear in mind 64bit version is way better than 32bit :)

You can try running yourself openssl if you have it:
openssl speed rsa dsa

There is no such thing as a 3.7ghz conroe at stock. But there is a 3ghz AMD you can buy. So are you saying from this benchmark your proving that Intel even with a 700mhz lead "In fantasy until the end of 2007" it hardly makes a difference to the 3ghz AMD? "However K8L will be out by then"

[doompc]

Looks like every key verify in this bench have to retrive data from system memory, so the IMC rule all the way.

[vitaminc]

1) brisbane is very popular right now with the OEMs (so you'll have to wait until they have excess stock)
2) K8L isn't due until 2H 2007. Please be patient. It would be kinda like me back January 2006 and :banana::banana::banana::banana::banana:ing about how conroe isn't out yet

1. AMD systems are not selling at the major OEMs, so the OEM inventories get returned to AMD and then reallocated to channels. Thus the retail pricing for AMD processors has gone down pretty fast in the last 2 weeks of December.
2. Correct. And K8L is something worth waiting for. :)

[Epsilon84]

There is no such thing as a 3.7ghz conroe at stock. But there is a 3ghz AMD you can buy. So are you saying from this benchmark your proving that Intel even with a 700mhz lead "In fantasy until the end of 2007" it hardly makes a difference to the 3ghz AMD? "However K8L will be out by then"

But it won't matter because we don't care about 10% performance difference. Everything is so fast already. Intel will be just as fast as K8L because we can't FEEL the difference. Software has to catch up to CPUs, it doesn't matter what you use.

Sound familiar Serge? LOL ;)

It's funny how AMD fanboys get excited about the smallest performance advantage, yet when the tables turn suddenly 'performance doesn't matter, it's already fast enough'. Oh, the hypocrisy! :nono:

[freeloader]

1. AMD systems are not selling at the major OEMs, so the OEM inventories get returned to AMD and then reallocated to channels. Thus the retail pricing for AMD processors has gone down pretty fast in the last 2 weeks of December.
2. Correct. And K8L is something worth waiting for. :)

No one knows what K8L is going to do. It'd be nice for AMD to leak some numbers, but hell will freeze over before that happens.

[Sparky]

Where is AMD? Relax, AMD had the performance crown for 3 years. Intel now has had it for not even a year yet and people are asking where AMD is? Intel can be ahead for a little while, it isn't that big a deal. I'm confident AMD will come back and bite intel ;)
And then intel will come back and bite AMD, and AMD bite intel, and intel bite AMD, and AMD bite intel.....
Never ending cycle :)

But I won't buy intel. I don't like the way they have conducted business. If they clean up their act then I won't have a problem buying from them but at this point I don't like 'em so I don't buy 'em. Besides, I've NEVER been on the cutting edge anyway :p:

[kiwi]

Did run again on the same OS, kernel version and openssl version. Results are shocking :D Venice @ 2.6Ghz easily beats conroe at stock 2.4ghz and is not so far behind conroe @ 3.6Ghz

Venice 2610Mhz, RAM value 512MB 200Mhz 2.5-3-3-7
http://www.xtremesystems.org/forums/attachment.php?attachmentid=54845&stc=1&d=1168961638

Conroe 2.4Ghz stock, RAM 2x1GB 400Mhz 4-4-4-8
http://www.xtremesystems.org/forums/attachment.php?attachmentid=54844&stc=1&d=1168961638

Conroe 9x400, RAM 2x1GB 400Mhz 4-4-4-8
http://www.xtremesystems.org/forums/attachment.php?attachmentid=54846&stc=1&d=1168961638


I have 3 possible explanations:

1. Software very optimized for amd cpus and their 64 bit extension
2. IMC helps a lot
3. EMT64 is not a true 64bit, need to do more research on this
Also look that Intel has 36 bits physical address size whereas AMD has 40 bits

I am referring to 64bit a lot because there is also a huge difference if you run it on 32bit win/linux and 64bit linux. Here you can see that 64bit finally makes a difference 2-3x times :)


Intel produces no consumer 64 bit chips.
The Prescott "J" model with the EMT64 designator is a Pentium 4 chip. Like
all other P4 chips, it is 32 bit. The EMT64 stands for Extended Memory
Technology, and allows the processor to use the x64 core coding to address
more than 4GB of RAM. It will emulate running the 64 bit code in Windows Xp
x64, but will run it in 32 bit mode only.
QUOTE
... the EMT64 chips are still 32 bit
chips. Intel will not have any 64 bit consumer chips until next year
(hopefully). EMT64 (Extended Memory Technology 64) chips allow a P4 chip to
use the same registers that a 64 bit chip accesses, so that they can use
more than the 4GB that a non-EMT64 P4 can access. As a side advantage of
this, the chip can "execute" the x86 code, but it does so through emulation
using a 32 bit thunking layer. It does not run the 64 bit OS as a 64 bit
OS, but as a thunked OS. This is slow and clumsy at best. The point I am
trying to make is that the EMT64 bit chips *ARE NOT* 64 bit chips. They are
able to use certain functions of the x86 extensions that allow it to access
the same registers, but that is all. It is *NOT* a 64 bit chip, not will it
ever be. Intel has announced that it *WILL* produce 64 bit chips next year.
I want those who are considering buying a 64 bit computer to understand that
while the EMT64 can emulate execution of the code, it is *NOT* a 64 bit
chip. Never was, never will be.

[Lightman]

Interesting findings ...

On the side note Prescott had additional ALU build in to deal with EMT64 overead. But it was still slower than proper implementation ;) .

Dunno how Intel added 64bit to Core2Duo....

[savantu]

The level of BS in the last 2 posts is astounding.

Not only is EM64T true implementation of 64bit , but the CPUs also have a 36bit physical adress space.

The difference between P4/Core and K8 in going from 32 to 64 is only a matter of implementation.It looks like some people forget that the internals of these cores vary tremendously.

As a result , depending on app some gain more or less from going to 64bit.Saying Intel somehow emulates 64bit is as stupid as saying the Earth is flat.But as history showed us , there are always people who take as truth what they want to be the truth.

[Lightman]

So you're denying that Prescott had 2 32bit ALU's??

Flaming is your way of discussion?

Short on memory?

Refreshment:

http://www.chip-architect.com/news/2003_04_20_Looking_at_Intels_Prescott_part2.html


Second integer core for 64 bit processing (not for multithreading)



It is as good as sure that the second 32 bit core is exclusively used for 64 bit processing, and in a way similar to the good old bit slices. There was the 4-bit AMD 2901 that could be used to build 16, 32 or 64 bit processors. The fact that makes it possible is because the core's is limited mainly to additive and logic functions. A 64 bit staggered addition will take a total of four 1/2 cycles but you can start two of them back to back on 1/2 cycle intervals. The latency to access the cache also does not need to be increased because of the extension to 64 (48) bit addresses. The higher part of the address is only used several cycles later to check the address tags with the TLB entries and not to access the data cache itself. What will increase with one cycle is the latency from an ALU instruction to a normal speed integer instructions. This delay will increase from 2 to 3 cycles. One extra pipeline stage is needed as well, resulting in a minor increase in the branch miss prediction penalty.



The reason that we can be so sure that the second core is not used to boost the 32 bit Hyper threading capabilities is the scheduler. This unit is by far the biggest entity on the Pentium 4 die. It is larger then all the Floating Point, MMX and SSE hardware together. It is not only big but it also consist mostly out of very timing critical optimized macro cells laid out by hand. It takes a lot of time and effort to change the scheduler. We've looked to it in detail and concluded that it has mainly remained unchanged on Prescott's die. This means that the maximum uOp throughput remains six per cycle using the same dispatch ports as the Pentium 4.


http://en.wikipedia.org/wiki/EM64T#Differences_between_AMD64_and_Intel_64


There are a small number of differences between each instruction set. Compilers generally produce binaries that target both AMD64 and Intel 64, making the differences mainly of interest to compiler developers and operating system developers.

[edit] Currently

* Intel 64's BSF and BSR instructions act differently when the source is 0 and the operand size is 32 bits. The processor sets the zero flag and leaves the upper 32 bits of the destination undefined.

* AMD64 supports 3DNow! instructions. This includes prefetch with the prefix 0F followed by opcode 0D and PREFETCHW, which are useful for hiding memory latency.

* Intel 64 lacks the ability to save and restore a reduced (and thus faster) version of the floating-point state (involving the FXSAVE and FXRSTOR instructions).

* Intel 64 lacks some model-specific registers that are considered architectural to AMD64. These include SYSCFG, TOP_MEM, and TOP_MEM2.

* Intel 64 supports microcode update as in 32-bit mode, whereas AMD64 processors use a different microcode update format and control MSRs.

* Intel 64's CPUID instruction is very vendor-specific, as is normal for x86-style processors.

* Intel 64 supports the MONITOR and MWAIT instructions, used by operating systems to better deal with Hyper-threading.

* AMD64 systems allow the use of the AGP aperture as an IOMMU. Operating systems can take advantage of this to let normal PCI devices DMA to memory above 4 GiB. Intel 64 systems require the use of bounce buffers, which are slower.

* SYSCALL and SYSRET are also only supported in IA-32e mode (not in compatibility mode) on Intel 64. SYSENTER and SYSEXIT are supported in both modes.

* Near branches with the 66H (operand size) prefix behave differently. One type of CPU clears only the top 32 bits, while the other type clears the top 48 bits.


and another one:

http://www.chip-architect.com/news/2002_04_16_Prescott_Prospects_.html#64%20bit%20Yam hill%20implementation%20may%20cost%20less%20then%2 02%20%%20extra%20die%20space

Basic 64 bit integer operations

A 64 bit extension by itself does not imply that the Integer Execution Unit and the Integer Register File have to be extended to 64 bit. A minimal implementation would simply use the 32 bit integer pipeline for 64 bit integer operations. The Floating Point/MMX/SSE pipelines are already 64 bit. No need for changes here.

The dual 'Rapid Execution' Units and the 32 bit register file run a twice the frequency and are together able to handle two 64 bit operations per cycle. (The Hammer is able to do 3 per cycle but its 64 bit additions might have twice the latency) The mechanisms to decode an operation into 2 sub-operations are already available in the pipeline. The 128 bit XMM/SSE operations for example are handled in two 64 bit pieces.

It would be advantageous if the basic functional timing of the rapid executions engines can remain the same. The current ones handle 32 bit additions as two skewed 16 bit ones. the 2nd addition starts 1/2 a cycle after the first when the carry bit is available. The newer integer ALU's seems to be fully 32 bit ALU's The same trick may thus be used to handle a 64 bit addition as two skewed 32 bit ones. Hardware for a full 32 bit addition takes about 15-20% longer as that for a 16 bit addition. It seems that Intel's circuit designers have closed this gap with novel design techniques like 'forward body biasing' et-cetera.

There is a lot more just GOOGLE for it...

[kiwi]

The level of BS in the last 2 posts is astounding.

Not only is EM64T true implementation of 64bit , but the CPUs also have a 36bit physical adress space.

The difference between P4/Core and K8 in going from 32 to 64 is only a matter of implementation.It looks like some people forget that the internals of these cores vary tremendously.

As a result , depending on app some gain more or less from going to 64bit.Saying Intel somehow emulates 64bit is as stupid as saying the Earth is flat.But as history showed us , there are always people who take as truth what they want to be the truth.

Maybe you can call BS my quote since I didn't come up with that anyway but perhaps you are so wise and explain why Conroe does so bad in this key generation?

[savantu]

So you're denying that Prescott had 2 32bit ALU's??

Flaming is your way of discussion?

Short on memory?

Refreshment:
....
There is a lot more just GOOGLE for it...

BS.

P4 used 2 16bit ALUs.Does that make it as "emulating" 32bit ?
It's all down to implementation which is something you fail to understand.

[Shintai]

Ehm...have people just for a short moment releaized that this is basicly the ONLY thing that shows such high difference? It could be like showing some sandra SSE benchmarks as a counter. But unless you are generating SSL keys day and night....

Also, it could also be its benchmark part of OpenSSL thats bugged. (It also seems to report bogus info)

Also the 64bit part on P4 is laughable. Its like saying K8 is emulating SSE, since it needs 2 cycles to do the 128bit instruction instead of 1.

And don´t mix 36bit PAE memory support with the x64 implementation.

The n00bieness of this 3rd page is amazing :o

[Lightman]

BS.

P4 used 2 16bit ALUs.Does that make it as "emulating" 32bit ?
It's all down to implementation which is something you fail to understand.

I've never used word 'emulating' for Intel EM64T...

I described P4 implementation of x64 as a slower one (my first post on this page), so how can it be that I fail to understand that??

:nono: