Sign in with
Sign up | Sign in
Your question

Conroe’s 64 bit performance – an Achilles heel!

Last response: in CPUs
Share
May 30, 2006 11:33:19 AM

This review shows that AM2 64 bit performance is typically much better than that of the Pentium D 8xx 9xx series. Has anyone tested a Conroe ES with 64 bit applications?
Surely Intel hasn’t overlooked this! They did drop the ball with the Pentium Pro, as it was heavily optimized for 32 bit and was very fast with Windows NT, but was actually slower than a Pentium with Windows 95, because 95 wasn’t a fully 32 bit O/S, too many 16 bit dependencies. A bit dumb considering NT was mainly a workstation & server O/S, so most PCs didn’t benefit from Pentium Pro’s excellent 32 bit performance.
This situation here is the reverse as they need to optimize it for the next generation O/S as it will be the mainstream one.

AM2 @ 65nm with Vista 64 sounds promising!

P.S. – I’m platform agnostic, I hope all the new chips are good. No fanboys please
May 30, 2006 12:40:08 PM

See the first thread in this forum.
May 30, 2006 12:41:06 PM

I didn't do any benchmarking but I didn't notice any difference between Vista Feb. CTP 32-bit and Vista Feb. CTP 64-bit on a 1.66GHz dual core Merom at work. Definitely see a performance difference though on a 3.0GHz P4 with HT between the OSs. Course the Merom system had 2GB and the P4 only had 1GB. Haven't seen how the 64-bit version acts with 2GB yet on the P4 since we recently upgraded it.

Vista Beta 2s out now so I'll installing that in not too long.
Related resources
May 30, 2006 12:47:24 PM

cool, very impressive.
vista will be 64-bit only?
May 30, 2006 12:47:42 PM

Quote:
This review shows that AM2 64 bit performance is typically much better than that of the Pentium D 8xx 9xx series. Has anyone tested a Conroe ES with 64 bit applications?
Surely Intel hasn’t overlooked this! They did drop the ball with the Pentium Pro, as it was heavily optimized for 32 bit and was very fast with Windows NT, but was actually slower than a Pentium with Windows 95, because 95 wasn’t a fully 32 bit O/S, too many 16 bit dependencies. A bit dumb considering NT was mainly a workstation & server O/S, so most PCs didn’t benefit from Pentium Pro’s excellent 32 bit performance.
This situation here is the reverse as they need to optimize it for the next generation O/S as it will be the mainstream one.


Up until very recently, Intel's 64-bit extensions, the EM64T lacked feature and performance wise to the AMD's 64-bit. They were talking about having updated steppings to add all the missing ones back in. Nobody seems to have tested that, but performance should improve.

Another point to make is that AMD's Athlon 64/X2 is faster at 32-bit already, doesn't it make sense that it will have a lead at 64-bit??? If you know that, good.

Preliminary benchmarks in 64 and 32-bit is that Woodcrest and Opteron benefits equally from 64-bit.
May 30, 2006 12:54:50 PM

Quote:
Conroe seems to be slower in 64-bit than 32-bit, but still outpacing all other processors in both

http://www.xtremesystems.org/forums/showthread.php?t=10...


Oh, so you know how X2's perform with 64-bit in 3dmark eh?? As far as I know, there is no testing except at XS. From initial 64-bit benchmarks, Athlon 64's were slower at 64-bit than 32-bit. So what??
May 30, 2006 12:56:02 PM

Vista will be available in both 32 & 64 -bit versions

i am not 100% certain but i think both binaries will be included on the same dvd disk in retail versions
May 30, 2006 1:03:23 PM

Quote:
Conroe seems to be slower in 64-bit than 32-bit, but still outpacing all other processors in both

http://www.xtremesystems.org/forums/showthread.php?t=10...


Oh, so you know how X2's perform with 64-bit in 3dmark eh?? As far as I know, there is no testing except at XS. From initial 64-bit benchmarks, Athlon 64's were slower at 64-bit than 32-bit. So what??

So.....i provided a link to some information showing a comparison bewteen conroe in 32 & 64 bit benchmarks, i never compared to X2. I'm not even really sure what you're asking!
May 30, 2006 1:05:00 PM

No need to worry crow, it appears that Conroe's 64-bit performance is quite good. Better then 32-bit in fact. Past Pentium systems would actually suffer from degraded 64-bit performance, but Conroe gains performance when moving to 64-bit as it should be.

CINEBENCH 2003 32-bit on Conroe: 642/1190
CINEBENCH 2003 64-bit on Conroe: 759/1401

These are at the same clockspeed and FSB/memory timings.

32-bit:



64-bit:



Fact suggest that Conroe 64-bit performance is great.

Cheers!
May 30, 2006 1:10:07 PM

hmmmmm i feel fanboyism in the air
May 30, 2006 1:17:04 PM

This article is very biased to AMD. They are promoting a general stereotypic conclusion on comparison between a low costs and low performance procesors from Intel(P-D 840 and P-D 940) and twice and more times expencive, top performance processors form AMD(s939 A64-FX-60, sAM2 A64-X2 5000+, sAM2 A64-FX-62)!
Only tests on which the K8 is performing the best and better than the Netburst, are selected: games, 3D Mark, SuperPi, Sisoft Sandra, ScienceMark, Maya, etc.
There are no Pentium 965XE, Pentium D 960 included in the comparison. Here I am skipping Core Duo, becouse it 32bit, no meetter that is performing better in 32bit mode than same clocked K8 dualcore. The P-XE is performing almost equal in average as the FX-60, but soon are coming new Intel chips with much efficient architecture significantly outperfroming everything clock-for-clok. Yo have mentioned in the Topic, but it is nowhere mentioned in the article, Intel Conroe (Core2 Duo)(see the Conroe thread on this subforum).

We all know that the Netburst iss unsucessfull. Intel finaly concluded their mistake after paying the price of the lost market share. They forgot the Netburst and made completly different architecture performing excelent in 64bit mode.
Anyway, the factor for choosing desktop and workstation is price/performance.
Intel prices:




AMD prices:

AMD Athlon 64 FX 62 - 2.8GHz - 64+64K L1 - 1MBx2 L2 - 125W - $1031
AMD Athlon 64 X2 5000+ - 2.6GHz - 64+64K L1 - 512KBx2 L2 - 89W - $696
AMD Athlon 64 X2 4800+ - 2.4GHz - 64+64K L1 - 1MBx2 L2 - 89W - $645
AMD Athlon 64 X2 4600+ - 2.4GHz - 64+64K L1 - 512KBx2 L2 - 89W - $558
AMD Athlon 64 X2 4400+ - 2.2GHz - 64+64K L1 - 1MBx2 L2 - 89W - $470
AMD Athlon 64 X2 4200+ - 2.2GHz - 64+64K L1 - 512KBx2 L2 - 89W - $365
AMD Athlon 64 X2 4000+ - 2.0GHz - 64+64K L1 - 1MBx2 L2 - 89W - $328
AMD Athlon 64 X2 3800+ - 2.0GHz - 64+64K L1 - 512KBx2 L2 - 89W - $303
AMD Athlon 64 3800+ - 2.4GHz - 64+64K L1 - 512KB L2 - 62W - $290
AMD Athlon 64 3500+ - 2.2GHz - 64+64K L1 - 512KB L2 - 62W - $189
AMD Sempron 3600+ - 2.0GHz - 64+64K L1 - 256KB L2 - 62W - $123
AMD Sempron 3500+ - 2.0GHz - 64+64K L1 - 128KB L2 - 62W - $109
AMD Sempron 3400+ - 1.8GHz - 64+64K L1 - 256KB L2 - 62W - $97
AMD Sempron 3200+ - 1.8GHz - 64+64K L1 - 128KB L2 - 62W - $87
AMD Sempron 3000+ - 1.6GHz - 64+64K L1 - 256KB L2 - 62W - $77

Energy Efficient:

AMD Athlon 64 X2 4800+ Energy Efficient - 65W - $671 - ($26 price premium)
AMD Athlon 64 X2 4600+ Energy Efficient - 65W - $601 - ($43 price premium)
AMD Athlon 64 X2 4400+ Energy Efficient - 65W - $514 - ($44 price premium)
AMD Athlon 64 X2 4200+ Energy Efficient - 65W - $417 - ($52 price premium)
AMD Athlon 64 X2 4000+ Energy Efficient - 65W - $353 - ($25 price premium)
AMD Athlon 64 X2 3800+ Energy Efficient - 65W - $323 - ($20 price premium)
AMD Athlon 64 X2 3800+ Energy Efficient Small Form Factor - 35W - $364 - ($61 price premium)
AMD Athlon 64 3500+ Energy Efficient Small Form Factor - 35W - $231 - ($42 price premium)
AMD Sempron 3400+ Energy Efficient Small Form Factor - 35W - $145 - ($48 price premium)
AMD Sempron 3200+ Energy Efficient Small Form Factor - 35W - $119 - ($32 price premium)
AMD Sempron 3000+ Energy Efficient Small Form Factor - 35W - $101 - ($24 price premium)

Notice the price of the P-D 940, it is $240 now, and in this biased article it is compared to a $1031 sAM FX-62.
Notice the price reduction on all Netburst CPUs, when Woodcrest(4 june, after 3 days from today) come and latter with Conroe(23 july). The Pentium D805 2.66GHz dualcore 2x1MB L2 will cost only $93. For $4 more I can buy Sempron sAM2 3400+ 1.8GHz 256kB L2 singlecore. What do you think who will perform better on multithread optimized software and OS(Vista).
Conroe 2.4GHz with DDR667 CL5 performs better(or in worst case equal) than sAM2 FX-62 2.8GHz DDR800 CL3. The Conroe 2.4GHz will cost $316 or less than third of the FX-62 price($1031 OEM). The DDR2-800 CL3 is twice as expencive as the DDR667 CL5.

The AMD sAM2 K8 systems will be just far more expencive than same performance Intel LGA775 Netburst/Core2 system. Intel will offer performance that AMD can't match for mobile, desktop and 2P-4core workstations/servers. The sAM2 is promising nothing, it offers nothing more than the s939 and s940. It is just a way to atract people to sell the old K8 in new package combined with the marketing(like the article you've provided) and shiny stickers with DDR2/sAM2/20.8GB/s total bandwith/64bit/HTT contents and logos.
May 30, 2006 1:18:43 PM

What do you make of the other 64-32bit comparisons though?
May 30, 2006 1:35:59 PM

Quote:
What do you make of the other 64-32bit comparisons though?
There are no others. Cinebench was the only application in that thread that is recompiled for 64-bit. The other examples were 32-bit apps running in 64-bit Windows. So they are not good examples of pure 64-bit performance.

There are other pure 64-bit workloads out there for reference, such as Linpack, certain SPEC_ints, etc.. but those are Woodcrest benchmarks and don't really apply to what most users here would care about.
May 30, 2006 1:52:58 PM

Quote:
No need to worry crow, it appears that Conroe's 64-bit performance is quite good. Better then 32-bit in fact. Past Pentium systems would actually suffer from degraded 64-bit performance, but Conroe gains performance when moving to 64-bit as it should be.

CINEBENCH 2003 32-bit on Conroe: 642/1190
CINEBENCH 2003 64-bit on Conroe: 759/1401
I appreciate that Conroe will also gain in the move to 64 bit, but my question is, ‘Does Athlon 64 gain more than Conroe in the move to 64 bit’, as my link in first post shows that it does versus Pentium D.

Quote:
This article is very biased to AMD. They are promoting a general stereotypic conclusion on comparison between a low costs and low performance procesors from Intel(P-D 840 and P-D 940) and twice and more times expencive, top performance processors form AMD(s939 A64-FX-60, sAM2 A64-X2 5000+, sAM2 A64-FX-62)!
Even if you are correct, you’ve missed the point here. The only comparison you need to make is Intel 32 bit versus Intel 64 bit & AMD 32 bit versus AMD 64 bit, which negates your perceived bias. The review clearly states in bold letters that the DivX encoder has been optimised for Athlon 64, so they aren’t exactly being underhand.
It shows that percentage wise AMD gains more from the move to 64 bit. Hence my question, has anybody benchmarked Conroe on XP-64 to see if they’ve tweaked the 64 bit support?

Quote:
Preliminary benchmarks in 64 and 32-bit is that Woodcrest and Opteron benefits equally from 64-bit.
Thanks, do you have a link for this comparison?

Quote:
Another point to make is that AMD's Athlon 64/X2 is faster at 32-bit already, doesn't it make sense that it will have a lead at 64-bit?
No. I’m wondering how Conroe compares against AM2 in 64 bit, it seems certain to me that Intel will win that one at 32 bit. So, I’m wondering if Intel has actually tweaked their 64 bit support in the Core 2 range of chips. Your quote above sounds as if they might well have.

Quote:
hmmmmm i feel fanboyism in the air
Not from me. I would prefer Intel to have sorted out their 64 bit performance issue, if they have one, so that I can move from an X2 3800+ to a Conroe or Merom. This is mainly on pricing as the AM2 EE (35W) X2 3800+ is currently a bit rich for me. But AM2 on 65nm also obviously has potential. My goal is good 64 bit performance per watt with a limit of ~45W and not ultimate clock speed, otherwise Conroe would walk it until possibly AM2 on 65nm.

I suppose another question is has MS tweaked the 64 bit performance in Vista compared to XP-64?
May 30, 2006 1:55:23 PM

i think the most impressive is the performace gain compared to the chip running in 32-bit, not when compared to intel's chip.
now this thread will become a neverending IntelxAMD war.

A platform versus a not released chip. It's the same thing as comparing an apple with an orange that is not released yet... pure speculation
May 30, 2006 2:07:05 PM

i wasnt referring to you :) 
i was referring to the guy who said: "No need to worry crow", like he was saying: hey, dont worry, our team will beat them hard, or something like that.

as for this
Quote:

Even if you are correct, you’ve missed the point here. The only comparison you need to make is Intel 32 bit versus Intel 64 bit & AMD 32 bit versus AMD 64 bit, which negates your perceived bias. The review clearly states in bold letters that the DivX encoder has been optimised for Athlon 64, so they aren’t exactly being underhand.
It shows that percentage wise AMD gains more from the move to 64 bit. Hence my question, has anybody benchmarked Conroe on XP-64 to see if they’ve tweaked the 64 bit support?

i guess we are postin at the same time, coz i said the same thing :p 
May 30, 2006 2:08:35 PM

Quote:
They did drop the ball with the Pentium Pro, as it was heavily optimized for 32 bit and was very fast with Windows NT, but was actually slower than a Pentium with Windows 95, because 95 wasn’t a fully 32 bit O/S, too many 16 bit dependencies. A bit dumb considering NT was mainly a workstation & server O/S, so most PCs didn’t benefit from Pentium Pro’s excellent 32 bit performance.


Mr. Fanboy, the Pentium Pro was designed for Windows NT and server environments.
May 30, 2006 2:23:34 PM

Quote:
Mr. Fanboy, the Pentium Pro was designed for Windows NT and server environments.
That’s interesting as I don’t remember reading that at the time. Are you absolutely sure that it was designed as a Workstation/Server chip and didn’t just end up being one by default :) 
Please drop the fanbay stance, it just makes you seem like a reactionary. Every time someone questions an aspect of a chip design, doesn’t automatically make them a fanboy. I was presenting hard data, which like any other data is open to interpretation and asking for opinions.
May 30, 2006 3:16:40 PM

Quote:
So.....i provided a link to some information showing a comparison bewteen conroe in 32 & 64 bit benchmarks, i never compared to X2. I'm not even really sure what you're asking!


I am saying it doesn't matter as there are apps which show all 64-bit capable CPUs being slower in 64-bit in certain apps. It's just that... another data. Sort of useless data since XS isn't comparing anything else. It's like showing SuperPI data between various Conroes, or various Yonah's for that matter. Unless they have a comparison, its useless. Fortunately its somewhat comparable for SuperPI, since it doesn't vary so much between different configurations as does some benchmarks.

Like dual core in games advantage. How Quake 4 gains a lot from dual core, while Call of Duty 2 loses performance. If only one CPU shows Call of Duty 2 performance increase with dual core, does it matter?? Not really...
May 30, 2006 3:34:56 PM

Quote:
iterations wrote:
No need to worry crow, it appears that Conroe's 64-bit performance is quite good. Better then 32-bit in fact. Past Pentium systems would actually suffer from degraded 64-bit performance, but Conroe gains performance when moving to 64-bit as it should be.

CINEBENCH 2003 32-bit on Conroe: 642/1190
CINEBENCH 2003 64-bit on Conroe: 759/1401

I appreciate that Conroe will also gain in the move to 64 bit, but my question is, ‘Does Athlon 64 gain more than Conroe in the move to 64 bit’, as my link in first post shows that it does versus Pentium D.


I actually have something to say for both of you. First, iterations: its difficult to judge performance gain on single application, especially cinebench, which shows that Pentium XE 840 is 15% faster than Pentium D 840, when in many cases Pentium D 840 can be faster than the XE. Therefore, Cinebench isn't that good of a benchmark. And of course, you can also find apps that Pentium D's gain in 64-bit.

Second, Crow: go to Tech-report for Woodcrest comparisons. Overall it gains similarly in 64-bit to Opteron(or loses similarly in 64-bit). 2CPU also has comparisons, I think.
Quote:

It shows that percentage wise AMD gains more from the move to 64 bit. Hence my question, has anybody benchmarked Conroe on XP-64 to see if they’ve tweaked the 64 bit support?


Tech-Report with Woodcrest, and 2CPU, also with Woodcrest, but with more benches.

http://www.2cpu.com/review.php?id=112

Actually 2CPU only shows that Woodcrest is still faster than Opteron in 64-bit. It doesn't seem to show a comparison.

Quote:
Mr. Fanboy, the Pentium Pro was designed for Windows NT and server environments.

That’s interesting as I don’t remember reading that at the time. Are you absolutely sure that it was designed as a Workstation/Server chip and didn’t just end up being one by default
Please drop the fanbay stance, it just makes you seem like a reactionary. Every time someone questions an aspect of a chip design, doesn’t automatically make them a fanboy. I was presenting hard data, which like any other data is open to interpretation and asking for opinions.


Can't really say whether Pentium Pro was made for workstation, but some claims say Intel just failed to put it at desktop. I'll just ignore the two sides.

It's not correct to refer back to Pentium Pro time to now. The situation is somewhat different. The one that looks same is really Itanium, with poor 32-bit performance but with excellent 64-bit performance.

(DrBlofeld, you do sound reactionary)
May 30, 2006 3:38:19 PM

Another point to note, at least in the referenced Mini-GZIP and DIVX Encoding benchmarks is that the 64-bit binaries are optimized specifically for AMD64 (as well as the 32-bit binary for DIVX encoding).

What I wonder is how much of a difference this makes? Going forward, do you guys think applications will be optimized specifically for AMD64, when a larger percentage of systems out there are still Intel-based? It seems unlikely to me.
May 30, 2006 3:41:34 PM

Quote:
i wasnt referring to you :) 
i was referring to the guy who said: "No need to worry crow", like he was saying: hey, dont worry, our team will beat them hard, or something like that.

I was the guy who said that but I think you misunderstood my intentions. I'm not an a "team", I'm merely a technology enthusiast, no matter which company makes the best tech etc...

Anyways, I said "don't worry" becuase it sounded like crow was worried that 64-bit performance was going to be the achilles heel for Conroe (see the title of the thread) but all of the facts available so far seem to suggest that Conroe performs very well in 64-bit task, better even than 32-bit. So then I provided the latest piece of hard evidence I had available.

That's all, it is simple, and there is no need for a battle, as some has suggested. We should all enjoy the competition and make intelligent decisions, using real data.

And DavidC1, you are absolutely right, CINEBENCH isn't the best example, it was just the only example I had at my fingertips. Thanks for the pointers to tech-report and 2cpu.
May 30, 2006 3:42:53 PM

Quote:
Another point to note, at least in the referenced Mini-GZIP and DIVX Encoding benchmarks is that the 64-bit binaries are optimized specifically for AMD64 (as well as the 32-bit binary for DIVX encoding).

What I wonder is how much of a difference this makes? Going forward, do you guys think applications will be optimized specifically for AMD64, when a larger percentage of systems out there are still Intel-based? It seems unlikely to me.


For the Pentium D's, Pentium 4's, Celeron D's before the patch that updates the EM64T to add the lacking instructions, its bad news.

For the ones after(which also includes Core), it won't matter, as they'll be one and the same.
May 30, 2006 4:16:24 PM

oh, okay sorry about that then

it's just because every intel versus amd topic becomes a war, and every fanboy, whether intel or amd is the most stupid and retarded person in the world. They simply forget that the main benefits of a good competition goes to the consumer, who gets better and cheaper products. Monopoly would end all that.
May 30, 2006 4:45:01 PM

hmmmmm
data can be manipulated, speculated, biased, fake... even worse from a processor that was not even oficially released

for me, until core is oficially released, any benchmark about it is not trusty
May 30, 2006 5:46:09 PM

It’s a real shame that two out of the four tests that they did used AMD optimized software. It seems a strange choice to me unless there aren’t versions of that software that aren’t AMD optimized, which would be unusual!

I thought the tests were worth posting because they’re the only ones that I’ve seen that ran 32/64 bit applications against 32/64 bit Windows on the same hardware platforms for both AMD & Intel. None of the links that others mentioned offered that level of direct comparison.
There’s certainly not enough there to even judge how Netburst compares with A64, but I thought it was food for thought. It’s possible that Conroe could even have a better 64 bit implementation than A64; I was just after some evidence.

I actually only posted it here because someone on another forum was trying to determine the most efficient platform for DivX encoding. When I noticed the substantial gains that 64 bit brought to encoding and that AMD gained a lot more than Intel, I was intrigued, hence this post. I hope there’s a decent amount of 64 bit applications & Codecs out by the time that Vista is released. Those benchmark gurus are gonna be in hog heaven :) 

Quote:
hmmmmm. data can be manipulated, speculated, biased, fake... even worse from a processor that was not even oficially released
for me, until core is oficially released, any benchmark about it is not trusty
Then why in the name of all that is holy in microprocessor design are you following this thread at all! It’s about Conroe; therefore it is bound to fit into some if not all of the categories that you seem to despise; are you a masochist?

Look, some people like to speculate for the sake of it, some because they are trying to decide whether to buy AM2 today or wait two months for Conroe, but ultimately it doesn’t matter, just please leave us alone to play in peace.
May 30, 2006 6:02:35 PM

because i've found some good info about AM2 performance, which is an oficially released processor :) 
May 30, 2006 6:50:51 PM

Even if the % increase in 64-bit apps vs. 32-bit for the Conroe isn't as good as AMD's, it will still outperform it's AMD counterparts.
May 30, 2006 6:53:20 PM

Quote:
Even if the % increase in 64-bit apps vs. 32-bit for the Conroe isn't as good as AMD's, it will still outperform it's AMD counterparts.


Are you 100% sure of this? :lol: 
May 30, 2006 7:08:00 PM

Quote:
because i've found some good info about AM2 performance, which is an oficially released processor :) 


jap0nes, I've got to say what good info. From what I've been seeing the performance gain in most cases is around 1% over 939 setups. I myself only own AMD products and have supported them for years. That doesn't mean however that I will only buy AMD. It is all about price/performance for me. Even though it hasn't been released yet I believe the chip to beat this go around is going to be the Conroe. In fact, I believe from this point on it is going to be a real battle between Intel and AMD. AMD is sure to come out with something to match the Conroe's performance in time, but when is the question and at what price. This alone concerns me because in such a contest where performance is equal Intel could have the advantage in price. Time will tell.
May 30, 2006 7:28:10 PM

Quote:
Even if the % increase in 64-bit apps vs. 32-bit for the Conroe isn't as good as AMD's, it will still outperform it's AMD counterparts.


Are you 100% sure of this? :lol: 

Do you still honestly believe that AMD will suddenly pull something amazing outof their ass, even though benchmarks OVER FOUR MEGS outperform AMD? that was the last cry for you AMD fanboys, until they ran every benchmark known to mankind and it clearly won EVERYTHING.
May 30, 2006 7:58:27 PM

i found interesting the performance gain from going 32 to 64-bit. In earlier benchmarks there were not much of an improvement, and i found a huge gain as new applications are ported to 64-bit.
Do i have to explain my thoughts now?
May 30, 2006 8:42:14 PM

You're right, the AM2 is going to stay just as it is, a re-packaged 939, no changes, what's there will always be there just like it is. AMD is promising nothing, just saying that they're working on new CPUs, but since you said nothing is going to change I guess you know more than they do.
Thanks for the look into the future.
May 30, 2006 8:55:33 PM

Quote:
i found interesting the performance gain from going 32 to 64-bit. In earlier benchmarks there were not much of an improvement, and i found a huge gain as new applications are ported to 64-bit.
Do i have to explain my thoughts now?


To be honest with you, no I don’t believe you have to explain your thoughts to me or to anyone else for that matter. Besides, I think I have a pretty good understanding of that already. I’m sorry that Conroe appears to be so good right now. In a couple of months hopefully it will be over one way or another and all our questions will be answered. My guess is Conroe is going to be a great chip. For some this is going to hurt deeply because they can’t stand the thought of AMD not being the performance leader. Is this you, I don’t know or care. Competition is good in any market and this is no different here. If nothing else it will force AMD to do something. In saying that, if Conroe is for real in July then I will have a very fast setup indeed at a lower price.
May 30, 2006 9:23:23 PM

You don't need to throw around the term "Fanboy", because at the moment there hasn't really been much to help prove whether or not Conroe's 64-bit performance will be that good.

Seeing as Intel hasn't mentioned much in their 64-bit performance, I'm thinking it may not be that great, but even with this small setback, the Conroe will probably outperform the AMD counterpart in 64-bit programs.
May 30, 2006 9:36:58 PM

Where you actually around when the Pentium Pro was released? I'm pretty darn sure that it was meant to be a server CPU by one thing alone: PRICE. I had (4) 486's in the house and was looking for a server when the Pro was released. Guess what, I couldn't afford that then, and I can't afford an Itanium now. Similiar differences between them and their mainstream counterparts.
May 30, 2006 10:05:32 PM

Quote:
Where you actually around when the Pentium Pro was released? I'm pretty darn sure that it was meant to be a server CPU by one thing alone: PRICE. I had (4) 486's in the house and was looking for a server when the Pro was released. Guess what, I couldn't afford that then, and I can't afford an Itanium now. Similiar differences between them and their mainstream counterparts.
Yes, I was around and the reviews I read suggested that it was a darn good 32 bit chip (NT) but was fatally flawed in it’s ability to run Windows 95. The fact is Intel killed Socket 8, which PP needed and didn’t develop the PP; I don’t think it ever ran above 200, whereas even the P1 ran at 233. So out came PII on Slot 1 which ramped up to 450 and supported SMP. The PP supported quad processors and larger caches so was a better server chip. But, if Socket 8 was always meant to be a standalone architecture for WS/Servers, why didn’t Intel continue to substantially develop it?

I don’t really know what the truth was on Intel’s intentions, so YES, I can’t back up my claim that Intel screwed up, but I don’t see anyone putting forward evidence to the contrary either. I honestly don’t care either way, but in the context of this post I feel it is useful information in that it shows how CPUs can perform very differently on different versions of Windows, which is what this thread is about in the first place.

If you feel so strongly about this I suggest you edit the Pentium Pro entry in Wikipedia, as they also seem to believe that Intel dropped the ball with the PP. Not that I’m suggesting that is a definitive source by any means.
May 30, 2006 10:08:45 PM

The pentium pro was awesome, the only problem was its cost.
May 30, 2006 11:41:10 PM

Quote:
This review shows that AM2 64 bit performance is typically much better than that of the Pentium D 8xx 9xx series. Has anyone tested a Conroe ES with 64 bit applications?
Surely Intel hasn’t overlooked this! They did drop the ball with the Pentium Pro, as it was heavily optimized for 32 bit and was very fast with Windows NT, but was actually slower than a Pentium with Windows 95, because 95 wasn’t a fully 32 bit O/S, too many 16 bit dependencies. A bit dumb considering NT was mainly a workstation & server O/S, so most PCs didn’t benefit from Pentium Pro’s excellent 32 bit performance.
This situation here is the reverse as they need to optimize it for the next generation O/S as it will be the mainstream one.


Up until very recently, Intel's 64-bit extensions, the EM64T lacked feature and performance wise to the AMD's 64-bit. They were talking about having updated steppings to add all the missing ones back in. Nobody seems to have tested that, but performance should improve.

Another point to make is that AMD's Athlon 64/X2 is faster at 32-bit already, doesn't it make sense that it will have a lead at 64-bit??? If you know that, good.

Preliminary benchmarks in 64 and 32-bit is that Woodcrest and Opteron benefits equally from 64-bit.

EMT64 and AMD64 are both x86-64. There were some variations in the final implementation of x86-64 on AMD's behalf, hence the x86-64 version of XP would not recognize an EMT64 processor because of the minor changes AMD did. With regards to the identical IA extensions the K8 was build with it in mind, including the additional registers, while Intel caught with their pants down had to slap an emulation (thunking) layer to the P4 allowing memory access greater than 4 gig. P4's emulated x86-64 hence the gross performance issues that have been seen.

Core2 appears to have a fully implemented x86-64 engine including the registers.
May 30, 2006 11:47:09 PM

Quote:
They did drop the ball with the Pentium Pro, as it was heavily optimized for 32 bit and was very fast with Windows NT, but was actually slower than a Pentium with Windows 95, because 95 wasn’t a fully 32 bit O/S, too many 16 bit dependencies. A bit dumb considering NT was mainly a workstation & server O/S, so most PCs didn’t benefit from Pentium Pro’s excellent 32 bit performance.


Mr. Fanboy, the Pentium Pro was designed for Windows NT and server environments.

Actually the Pentium Pro was designed to be a 32bit processor in and out, while NT was a designed from the ground up on 32bit code for stability sake. MS knew that the home user wouldn’t care if 95 was flakey but network administrators would sure care if NT was.
May 31, 2006 12:08:06 AM

Quote:
Quote:
This review shows that AM2 64 bit performance is typically much better than that of the Pentium D 8xx 9xx series. Has anyone tested a Conroe ES with 64 bit applications?
Surely Intel hasn’t overlooked this! They did drop the ball with the Pentium Pro, as it was heavily optimized for 32 bit and was very fast with Windows NT, but was actually slower than a Pentium with Windows 95, because 95 wasn’t a fully 32 bit O/S, too many 16 bit dependencies. A bit dumb considering NT was mainly a workstation & server O/S, so most PCs didn’t benefit from Pentium Pro’s excellent 32 bit performance.
This situation here is the reverse as they need to optimize it for the next generation O/S as it will be the mainstream one.


Up until very recently, Intel's 64-bit extensions, the EM64T lacked feature and performance wise to the AMD's 64-bit. They were talking about having updated steppings to add all the missing ones back in. Nobody seems to have tested that, but performance should improve.

Another point to make is that AMD's Athlon 64/X2 is faster at 32-bit already, doesn't it make sense that it will have a lead at 64-bit??? If you know that, good.

Preliminary benchmarks in 64 and 32-bit is that Woodcrest and Opteron benefits equally from 64-bit.

EMT64 and AMD64 are both x86-64. There were some variations in the final implementation of x86-64 on AMD's behalf, hence the x86-64 version of XP would not recognize an EMT64 processor because of the minor changes AMD did. With regards to the identical IA extensions the K8 was build with it in mind, including the additional registers, while Intel caught with their pants down had to slap an emulation (thunking) layer to the P4 allowing memory access greater than 4 gig. P4's emulated x86-64 hence the gross performance issues that have been seen.

Core2 appears to have a fully implemented x86-64 engine including the registers.[/quote]

AMD has been working on an even better x86-64 engine for 4 years :wink:
May 31, 2006 12:35:23 AM

Quote:
Even if the % increase in 64-bit apps vs. 32-bit for the Conroe isn't as good as AMD's, it will still outperform it's AMD counterparts.


Are you 100% sure of this? :lol: 

Actually, relative % increase in 64-but vs 32-bit apps says nothing one way or the other about relative comparison between the processors.

If you get a 10% gain on 100 vs a 20% gain on 90, as in case A, then yes, the statement is true. If you get a 10% gain on 90 and a 20% on 100 though it is not true.

I guess what I am saying is why don't all of you people out on this forum that love to make unsubstantiated claims based on skewed facts and Inquirer articles please just SHUT UP and let it go.

Who cares if Company A beats Company B for 18 months or whatever? How does it affect the quality of your life?
May 31, 2006 1:05:03 AM

Quote:
Conroe seems to be slower in 64-bit than 32-bit, but still outpacing all other processors in both

http://www.xtremesystems.org/forums/showthread.php?t=101344
Someone didn't seem to be careful enougn to check to see that all the benchmarks are in 32 bit mode. Those tests are totally worthless. They are not 64 bit benchmarks. It is like taking windows 95 and running 16 bit benchmarks to test whether the PII was as good as the Pentium Pro. There is no way that the comparasion is valid. Before any one smarts off, I am a retired IEEE engineer certified in 1975 and worked first on ASCI Red(9800 Pentium pros) then Red Strom and the Dell Xeon EM64t machine. EM64t is good in 32 bit up to the 4gb memory barrier after that it chokes. That is what you have to look forward to in Conroe. Conroe's instruction decoding structure is 3 simple instructions and 1 complex instruction . AMD64 is 3 complex instructions. In 32 bit mode Conroe should blaze past AMD64. But come time to load and operate 64 bit software in 64 bit mode the three simple instruction decoders are going to be useless probalby about 75% of the time. That means in serious 64 bit computation Intel has 1 effective instruction decoder compared to 3 for AMD. We are on the cusp of a change that is nearly identical to the changeover from the 80286 16 bit chip to the 80386 32 bit chip. Both AMD64 and EM64T are really 32 bit cores that use 64 bit emulation instructions to process 64 bit code. The AM64 extensions do not have the limits of 4 gb of memory when running 32 bit code. Passages from several good articles that are simple enough to be understood by lay readers follow:

Some 64-bit processor architectures can execute 32-bit code natively without any performance penalty, such as AMD64, MIPS64,Sparc64, zSeries, PowerPC64, etc. This kind of support is commonly called biarch support or more generally multi-arch support.

A completely distinct change in microprocessor design since approximately 2004 has been the introduction of multicore (initially, dual core) processors, which include two or more separate processor cores in a single "chip" or package. Effectively, dual-core processors deliver two-way Symmetric multiprocessing computers using a single motherboard processor socket - or four-way machines with two sockets, and so on. Some dual-core x86 chips, such as AMD's Athlon 64 X2 range, are 64-bit capable; others, such as Intel's Core Duo, are 32-bit devices. The performance and capacity implications of multi-core processors are entirely independent and different from those of 64-bit versus 32-bit. A processor with two 32-bit cores is not a 64-bit processor.

http://en.wikipedia.org/wiki/64_bit

for Intel and the EM64t mode
CPUs with this technology have a new operating mode, called IA32E, which has two sub-modes:

* Compatibility mode: Allows 64-bit operating systems to run 32-bit and 16-bit software unmodified. The operating system can have 64-bit (in 64-bit mode), 32-bit and 16-bit (both on compatibility mode) programs running at the same time. However, 32-bit programs will run as if they were running on a 32-bit CPU, i.e. will access only up to 4 GB of RAM. The same idea goes to 16-bit program, that will still access only up to 1 MB of RAM.
* 64-bit mode: Allows 64-bit operation systems to use the new 64-bit addressing space provided by this technology.

As you can see above, EM64T technology can only be used by 64-bit operating systems, like the forthcoming Windows 64. 32-bit operating systems, like Windows XP, will continue to run under regular IA32 mode, i.e. using the regular 32-bit addressing space � in order words, can access only up to 4 GB RAM.

In plain English, Windows nowadays cannot take advantage of EM64T technology since it is still a 32-bit operating system.
http://www.hardwaresecrets.com/article/262
for AM64:
We start with a question: how to make the transition from the 32-bit CPUs to the 64-bit one? AMD is answering this question with an architecture that, besides the 64-bit environment, promises compatibility with al the programs developed to 16 and 32 bits. The aim is to offer a low cost solution for the users to make this transition in a very easy way. With an architecture compatible to the x86 world, the board and software manufacturers and the users can manage their investments more easily. The idea is to offer a secure bridge for the transition from 32 to 64 bits. The 64 -computing is directed to applications that are very hungry for memory, such as the great databases, the CAD tools and the simulations that, according to the present features, are limited by the 4 GB address space.

AMD is calling its new architecture x86-64 and it will be started with a family of processors that have the code name Hammer (the first project, code-named clawhammer, was released as Athlon 64). The 64-bit strategy by AMD is the extension of the present x86 CPUs to work at 64 bits, with the introduction of the so called Long Mode. This solution is safe because it has already been employed at the time of the transition from 16 bits (8088 and 286 CPUs) to 32 bits (386 CPUs and forward). Since long ago, the 32-bit CPUs operate in two modes. When in real mode, they become like the old 8088, but, when in protected mode, they offer 32-bit features, with task and memory managers. The x86-64 architecture offers a new mode called Long Mode, which serves for setting the CPU to operate at 64 bits. When in long mode, besides the 64-bit features, registers extended to 64 bits are offered and, besides that, new registers have been added. Let's go to the study of this new mode.
http://www.hardwaresecrets.com/article/56

I would strongly suggest that everyone read all these three articles and then you will have a basic grasp of what is happening. Intel is going to be better in the 32 bit mode but suffer significantly is 64 bit mode. That problem is abundantly clear to Dell and that is why they are offering 8XX opteron servers. Today buying a Conroe is like buying a 80286 cpu in 1989 with the 80386 looking down your throat.
May 31, 2006 2:13:54 AM

Quote:
Who cares if Company A beats Company B for 18 months or whatever? How does it affect the quality of your life?
If Intel has the better core, global warming will cause the icecaps to melt destroying the world as we know it. :wink:
May 31, 2006 6:48:59 AM

Oh I see!
so you think its not fair comparison to bench existing CPU's, to existing CPU's? the one's you and I can buy right now (not ES). Do you see it fair to compare an existing w/ conroe(non existing in market)?
how about both side be fair and test existing w/ existing?
Or lets wait till July 23rd when conroe is released and compare it w/ AMD's 65/45nm ES? would that be fair enough?
Or are you then, going to say "not fair to compare existing w/ non existing" ?
Yes It is possible that conroe may beat AM2's, but will it when AMD's 65nm are out?
Frankly for a 90nm doing 35watts is a quite accomplishment, can Intel do the same w/ 65nm? no. Intel's 65nm TDP is higher than AMD's s939, how is that for comparison?
Gota love IMC
May 31, 2006 7:26:24 AM

LMM your efforts to play down conroe with stupid reasoning are quite funny.

Quote:
Frankly for a 90nm doing 35watts is a quite accomplishment


Wow the 3800 is 35 watts, too bad it'll get owned by conroe.

Quote:
so you think its not fair comparison to bench existing CPU's, to existing CPU's? the one's you and I can buy right now (not ES). Do you see it fair to compare an existing w/ conroe(non existing in market)?


It comes out in a month. AM2 reviews came out before it was released. I guess we should have discounted those reviews.

Quote:
Or lets wait till July 23rd when conroe is released and compare it w/ AMD's 65/45nm ES? would that be fair enough?


What changes between now and then? Nothing.

Give up LMM and get a life.
May 31, 2006 7:49:46 AM

Quote:
What is it that 65 nm brings to AM2 that will get them to performance parity with the data we have seen to date?

Thanks,
Jack

All I care is: Today I can buy sAM2 that outperforms existing of anything, and much cheaper (CPU/MOBO combo less than $250) and give lots of amenities(check Nvedia's 590). above all, I will be able to upgrade to 65/45nm system without throwing away my mobo. I also know I can overclock the hell of DDR2 (Gota love IMC) and cpu w/ minimum power consumption. Price/ performance, cost of operation, overclocking, SLI, EPP, Faster ethernet/internet, 6 sataII, etc..etc...
what else do I want?
I am glad Intel pulled it off (finally), but so far AMD has provided me the right equipment for the right price. AMD has never failed me and I know it never will. That’s all I care.
You go have your WR and sweat it out, and I just sit here compute w/ my cool 35 watts A64 3500, and call it even.
!