Pentium M with Asus 478 adapter is very impressive.

[Keysplayr]

Originally posted by: Technonut
Posted @ Legit Reviews Forum:

I got some really bad news for the almost 10,000 viewers of this thread... The CT-479 card died today (3-31-2005). All i get is a system beep when using the card. I tried both the P4P800SE and my P4C800ED without any luck. Changed out the CPU, Processor, Video Card, Memory, put back on the stock cooling unit and still no luck. The System boots, has the system beep, then hangs before the post screen and beeps. The video card fan never kicks on and the keyboard is never detected. Since all the parts work fine on other systems the only thing left is the adapter. Could I have fried the adapter after running the system at 1.6V and 200MHz FSB?

RIP.....

Geez... Overclocking has become so commonplace that people think its abnormal not to do it. LOL. I'm sure if this guy hadn't o/c'd the part with higher voltage, he would still be using it safe and sound. Asus makes pretty decent components. He did push it pretty hard though. I'm sure ASUS is sending him another one so he can complete his review. (And I don't mean through warranty.. )

 

[DAPUNISHER]

Originally posted by: keysplayr2003
Your talking about current HT technology I assume. But what about tomorrow's HT? Will it require a lengthy pipline? We don't know that.

That is what I'm talking about. Presuming it will be implemented in the same fashion it is with a 31 stage pipe, I'd agree. But I don't think the purpose SMT will serve on the new stronger IPC CPUs will be identical. That .pdf I linked showed they managed some very nice gains with parallel SIMD and SMP in their one use for it. Since Intel is evidently showing some focus on SIMD, the use of SMT and CMP with parallel SIMD instructions could be why HT will stick around, couldn't it?

 

[classy]

Originally posted by: Frown66
New article on Pentium M on the ASUS adapter. They got it to 2.7ghz. Its a great gaming cpu:

http://www.gamepc.com/labs/view_content.asp?id=770ct479&page=8

I'm just curious what this cpu (overclocking-wise) can do with a better cooling solution. That stock HSF for it looks rather wimpy.

You gotta be careful with Gamepc because they can mislead people. They run all the A64 memory at cas 333 because they say its because its not fair to the Dothan because it can only utilize ddr on a slower fsb. Yeah OK :roll: Thats why those marks look so impressive. Anand was running @2.56 and at that speed only "equaled" the A64 in a few benchmarks. So I question the validity of the marks of Gamepc. And most of all the other reviews have have shown similar results to Anandtech. Gamepc is the only odd ball with huge overclocks.

 

[Keysplayr]

Originally posted by: DAPUNISHER

Originally posted by: keysplayr2003
Your talking about current HT technology I assume. But what about tomorrow's HT? Will it require a lengthy pipline? We don't know that.

That is what I'm talking about. Presuming it will be implemented in the same fashion it is with a 31 stage pipe, I'd agree. But I don't think the purpose SMT will serve on the new stronger IPC CPUs will be identical. That .pdf I linked showed they managed some very nice gains with parallel SIMD and SMP in their one use for it. Since Intel is evidently showing some focus on SIMD, the use of SMT and CMP with parallel SIMD instructions could be why HT will stick around, couldn't it?

I absolutely agree. I hope that HT will stick around, but not in the form of a crutch. I would like it as an added bonus to an already highly capable processor with HT turned off.
Then again, what I would like, and what I would get are two different things aren't they..... hehe.

 

[DAPUNISHER]

The potential problem with gamepc is that they sell the stuff

 

[Keysplayr]

Originally posted by: classy

Originally posted by: Frown66
New article on Pentium M on the ASUS adapter. They got it to 2.7ghz. Its a great gaming cpu:

http://www.gamepc.com/labs/view_content.asp?id=770ct479&page=8

I'm just curious what this cpu (overclocking-wise) can do with a better cooling solution. That stock HSF for it looks rather wimpy.

You gotta be careful with Gamepc because they can mislead people. They run all the A64 memory at cas 333 because they say its because its not fair to the Dothan because it can only utilize ddr on a slower fsb. Yeah OK :roll: Thats why those marks look so impressive. Anand was running @2.56 and at that speed only "equaled" the A64 in a few benchmarks. So I question the validity of the marks of Gamepc. And most of all the other reviews have have shown similar results to Anandtech. Gamepc is the only odd ball with huge overclocks.

Intel Pentium-M System Configurations
Processor(s) Intel Pentium-M 770 (2.13 GHz), 32k L1, 2048k L2 Cache, 533 MHz FSB
Intel Pentium-M 755 (2.00 GHz), 32k L1, 2048k L2 Cache, 400 MHz FSB

Memory 2 x Samsung DDR-400 (PC-3200) Memory - 1 GB Total
CAS 2,2,2 Latency at DDR-333 Speeds
Run at DDR-333 Speeds Due To Motherboard / Chipset Limitations

Motherboard Asus P4P800-SE Motherboard with Asus CT-479 Adapter
Intel 865PE Express Chipset
Intel 6.3.0 Chipset Drivers Installed

DFI 855GME-MGF Motherboard
Intel 855GME Chipset
Intel 6.3.0 Chipset Drivers Installed


Graphics Card nVidia GeForce 6800 Ultra 256MB AGP 8x
nVidia Forceware 71.84 Drivers Installed

Hard Disk Western Digital Raptor 74GB - 10,000 RPM - 8MB Buffer - Serial ATA/150

Operating System Windows XP Professional Edition - Default Install
Windows XP Service Pack 2 and Direct X 9.0C Installed


AMD Athlon64 / FX System Configurations
Processor(s) AMD Athlon64 FX-55 (2.6 GHz) - 128k L1, 1024k L2 Cache
AMD Athlon64 4000+ (2.4 GHz) - 128k L1, 1024k L2 Cache
AMD Athlon64 3800+ (2.4 GHz) - 128k L1, 512k L2 Cache
AMD Athlon64 3500+ (2.2 GHz) - 128k L1, 512k L2 Cache
AMD Athlon64 3200+ (2.0 GHz) - 128k L1, 512k L2 Cache

Memory 2 x Samsung DDR-400 (PC-3200) Memory - 1 GB Total
CAS 3,3,3 Latency at DDR-400 Speeds

Motherboard Asus A8N-SLI Deluxe Motherboard
nVidia nForce4-SLI Chipset
nVidia Forceware 6.53 Chipset Drivers Installed

Graphics Card nVidia GeForce 6800 Ultra 256MB PCI Express x16
nVidia Forceware 71.84 Drivers Installed

Hard Disk Western Digital Raptor 74GB - 10,000 RPM - 8MB Buffer - Serial ATA/150

Operating System Windows XP Professional Edition - Default Install
Windows XP Service Pack 2 and Direct X 9.0C Installed


Seems like the Athlons are running at DDR400 to me. And the Pentium M's are running at DDR333 due to motherboard/chipset limitations.

So either I misunderstood you, or you misunderstood the TestBed Notes. So how much of a performance loss does DDR400 at CAS 3.3.3 take over DDR333 at 2.2.2? Probably evens out I would think. I'm no overclocker, so can someone either credit, or discredit this claim that the Athlons were "handicapped" in the GamePC review? Thanks.

 

[pm]

Originally posted by: clarkey01
To back Duive up he's saying that the smaller pipeline in Yonah and Presler will not benifit from HT. Yes G5 use a form of HT and there only 17 or so pipelines deep, an Athlon 64 doesnt require or would benifit greatly from HT.

The CPU that I work on, Montecito, - an upcoming member of the Itanium family - has a fairly short pipeline and implements MT on both of it's cores. We see a substantial boost in some applications - particularly transactional computing.

 

[DAPUNISHER]

Originally posted by: keysplayr2003
I absolutely agree. I hope that HT will stick around, but not in the form of a crutch. I would like it as an added bonus to an already highly capable processor with HT turned off.
Then again, what I would like, and what I would get are two different things aren't they..... hehe.

Anand says we will see more SMT not less "4 threads per core" I don't think they would effort this if they couldn't make HT a strong element on a shorter piped CPU.

 

[DAPUNISHER]

Originally posted by: pm

Originally posted by: clarkey01
To back Duive up he's saying that the smaller pipeline in Yonah and Presler will not benifit from HT. Yes G5 use a form of HT and there only 17 or so pipelines deep, an Athlon 64 doesnt require or would benifit greatly from HT.

The CPU that I work on, Montecito, - an upcoming member of the Itanium family - has a fairly short pipeline and implements MT on both of it's cores. We see a substantial boost in some applications - particularly transactional computing.

Thanks for joining in pm :beer: I was hoping some of you guys would at least say something that didn't compromise your employment. Am I way off in left field with the parallel SIMD? can you even answer yes or no?

 

[Keysplayr]

Originally posted by: DAPUNISHER

Originally posted by: keysplayr2003
I absolutely agree. I hope that HT will stick around, but not in the form of a crutch. I would like it as an added bonus to an already highly capable processor with HT turned off.
Then again, what I would like, and what I would get are two different things aren't they..... hehe.

Anand says we will see more SMT not less "4 threads per core" I don't think they would effort this if they couldn't make HT a strong element on a shorter piped CPU.

This may mean that "Longhorn" for the desktop will not have a 2-processor limitation as NT/2K/XP currently does for the desktop. Whether they be logical or physical. Just a guess. It would be pretty cool if quad or more processors were not indigenous to Windows server operating systems.

 

[DAPUNISHER]

Windows XP professional supports 2 physical processors or 2 physical+2 logical which I think it treats as 4 logical processors braddah.

 

[imported_michaelpatrick33]

Originally posted by: keysplayr2003

Originally posted by: classy

Originally posted by: Frown66
New article on Pentium M on the ASUS adapter. They got it to 2.7ghz. Its a great gaming cpu:

http://www.gamepc.com/labs/view_content.asp?id=770ct479&page=8

I'm just curious what this cpu (overclocking-wise) can do with a better cooling solution. That stock HSF for it looks rather wimpy.

You gotta be careful with Gamepc because they can mislead people. They run all the A64 memory at cas 333 because they say its because its not fair to the Dothan because it can only utilize ddr on a slower fsb. Yeah OK :roll: Thats why those marks look so impressive. Anand was running @2.56 and at that speed only "equaled" the A64 in a few benchmarks. So I question the validity of the marks of Gamepc. And most of all the other reviews have have shown similar results to Anandtech. Gamepc is the only odd ball with huge overclocks.

Intel Pentium-M System Configurations
Processor(s) Intel Pentium-M 770 (2.13 GHz), 32k L1, 2048k L2 Cache, 533 MHz FSB
Intel Pentium-M 755 (2.00 GHz), 32k L1, 2048k L2 Cache, 400 MHz FSB

Memory 2 x Samsung DDR-400 (PC-3200) Memory - 1 GB Total
CAS 2,2,2 Latency at DDR-333 Speeds
Run at DDR-333 Speeds Due To Motherboard / Chipset Limitations

Motherboard Asus P4P800-SE Motherboard with Asus CT-479 Adapter
Intel 865PE Express Chipset
Intel 6.3.0 Chipset Drivers Installed

DFI 855GME-MGF Motherboard
Intel 855GME Chipset
Intel 6.3.0 Chipset Drivers Installed


Graphics Card nVidia GeForce 6800 Ultra 256MB AGP 8x
nVidia Forceware 71.84 Drivers Installed

Hard Disk Western Digital Raptor 74GB - 10,000 RPM - 8MB Buffer - Serial ATA/150

Operating System Windows XP Professional Edition - Default Install
Windows XP Service Pack 2 and Direct X 9.0C Installed


AMD Athlon64 / FX System Configurations
Processor(s) AMD Athlon64 FX-55 (2.6 GHz) - 128k L1, 1024k L2 Cache
AMD Athlon64 4000+ (2.4 GHz) - 128k L1, 1024k L2 Cache
AMD Athlon64 3800+ (2.4 GHz) - 128k L1, 512k L2 Cache
AMD Athlon64 3500+ (2.2 GHz) - 128k L1, 512k L2 Cache
AMD Athlon64 3200+ (2.0 GHz) - 128k L1, 512k L2 Cache

Memory 2 x Samsung DDR-400 (PC-3200) Memory - 1 GB Total
CAS 3,3,3 Latency at DDR-400 Speeds

Motherboard Asus A8N-SLI Deluxe Motherboard
nVidia nForce4-SLI Chipset
nVidia Forceware 6.53 Chipset Drivers Installed

Graphics Card nVidia GeForce 6800 Ultra 256MB PCI Express x16
nVidia Forceware 71.84 Drivers Installed

Hard Disk Western Digital Raptor 74GB - 10,000 RPM - 8MB Buffer - Serial ATA/150

Operating System Windows XP Professional Edition - Default Install
Windows XP Service Pack 2 and Direct X 9.0C Installed


Seems like the Athlons are running at DDR400 to me. And the Pentium M's are running at DDR333 due to motherboard/chipset limitations.

So either I misunderstood you, or you misunderstood the TestBed Notes. So how much of a performance loss does DDR400 at CAS 3.3.3 take over DDR333 at 2.2.2? Probably evens out I would think. I'm no overclocker, so can someone either credit, or discredit this claim that the Athlons were "handicapped" in the GamePC review? Thanks.

3-3-3 on the AMD 64. Who are they trying to kid? No one runs that high latency crap on the AMD 64. Additionally, I would be curious whether they have 1t or 2t command timings on the AMD 64. They don't state it, and if they set a 2t rate than the AMD would be 10-12% slower.

If AMD releases their dual core at 2.4 it will destroy Intel's 3.2 dual core as far as performance and thermal viability (I believe the dual core AMD's are 95w or 110w maxiumum wherease the Intel is 135w and of course AMD's is maximum rate [i don't know the technical terms, lol] while Intel's is an average rate). A FX-53/4000+ versus a Prescott 3.2 in single cpu performance is not even close and AMD's dual core actually sounds like truer "dual core" implementation than Intel's.

A dual core 2.4 AMD64 versus a dual core Dothan based 2.4 would be fun though!

 

[Ackmed]

Running 3-3-3 is idiotic, and I can think of only one reason why they would do that. To make the A64 slower.

 

[Keysplayr]

Originally posted by: Ackmed
Running 3-3-3 is idiotic, and I can think of only one reason why they would do that. To make the A64 slower.

Umm, nowhere in that article did anyone say they intentionally made the Athlon systems slower on purpose to level the playing field for the Pentium M platform. If I am wrong please show me. Its just starting to sound like reaching to me when someone grasps for little excuses as to why a processor does, or does not do so well.
So, how much of a difference is there from DDR333 @ 2.2.2 then DDR400 @ 3.3.3?
This is the question that will make or break the claim. Maybe.

Can we please make it our business to damper the flames (if any) in this thread.
So, lets take care of business folks. Thanks.

 

[Keysplayr]

Originally posted by: michaelpatrick33

Originally posted by: keysplayr2003

Originally posted by: classy

Originally posted by: Frown66
New article on Pentium M on the ASUS adapter. They got it to 2.7ghz. Its a great gaming cpu:

http://www.gamepc.com/labs/view_content.asp?id=770ct479&page=8

I'm just curious what this cpu (overclocking-wise) can do with a better cooling solution. That stock HSF for it looks rather wimpy.

You gotta be careful with Gamepc because they can mislead people. They run all the A64 memory at cas 333 because they say its because its not fair to the Dothan because it can only utilize ddr on a slower fsb. Yeah OK :roll: Thats why those marks look so impressive. Anand was running @2.56 and at that speed only "equaled" the A64 in a few benchmarks. So I question the validity of the marks of Gamepc. And most of all the other reviews have have shown similar results to Anandtech. Gamepc is the only odd ball with huge overclocks.

3-3-3 on the AMD 64. Who are they trying to kid? No one runs that high latency crap on the AMD 64. Additionally, I would be curious whether they have 1t or 2t command timings on the AMD 64. They don't state it, and if they set a 2t rate than the AMD would be 10-12% slower.

If AMD releases their dual core at 2.4 it will destroy Intel's 3.2 dual core as far as performance and thermal viability (I believe the dual core AMD's are 95w or 110w maxiumum wherease the Intel is 135w and of course AMD's is maximum rate [i don't know the technical terms, lol] while Intel's is an average rate). A FX-53/4000+ versus a Prescott 3.2 in single cpu performance is not even close and AMD's dual core actually sounds like truer "dual core" implementation than Intel's.

A dual core 2.4 AMD64 versus a dual core Dothan based 2.4 would be fun though!

Well, I see a lot of numbers and not much else. If your going to discredit something, make it undeniable. Get some facts together to disprove this article if you can.
Then we can safely say, "The Athlon Platform was crippled" which I doubt very much.

 

[VirtualLarry]

Originally posted by: keysplayr2003
Your talking about current HT technology I assume. But what about tomorrow's HT? Will it require a lengthy pipline? We don't know that. I believe Windows XPpro can only utilize up to 2 processors anyway, be it 2 physical CPUs or one Physical and one logical. Having HT on a dual core CPU would not benefit as XP will only use the "first logical" CPU Per core for a maximum of two CPU's. At least for now. Who knows what Longhorn will bring.

Actually, for licensing purposes, XP recognizes a HT-capable CPU as a single physical CPU. For scheduling purposes, it sees them as two logical CPUs. That's the primary difference between XP's and W2K's support for HT-capable CPUs. (IOW, XP Pro can use two HT CPUs, or a dual-core (each HT-enabled) CPU, because XP Pro is licensed for two CPUs, and it sees HT-capable cores as a single CPU. W2K Pro is also licensed for two CPUs, but to it, an HT-capable CPU core appears as two SMP CPUs. But because of the recommended default CPU MPS table layout, the first two physical CPU cores are listed first, then the two logical ones, so W2K Pro would actually schedule threads based on the two seperate physical CPUs or cores in a dual-core CPU, so that effectively HT would be disabled in that scenario, unless you moved up to W2K Server/Advanced Server.)

 

[VirtualLarry]

Originally posted by: pm
The CPU that I work on, Montecito, - an upcoming member of the Itanium family - has a fairly short pipeline and implements MT on both of it's cores. We see a substantial boost in some applications - particularly transactional computing.

Which is likely why Intel isn't killing HT - because SMT can be useful in some server scenarios as you mentioned, and I'm sure Intel wants to compete against Sun's Niagara chip in the TP/blade space.

 

[xtknight]

Originally posted by: Sentential

Originally posted by: geforcetony

Originally posted by: BadThad
Clock for clock, the Pentium M is a VERY strong cpu. The Asus adapter is pretty cool, but this CPU really needs a dedicated platform to reach it's full desktop potential IMO.

The P-M can use whatever memory the chipset manufacturers want,

Thats false bud. When you change RAM type you also need to change the sense pins of the CPU. Intel didnt go LGA775 for nothing ya know. Just for reference. More pins on the RAM dimms the more pins you need on the CPU.

AMD's DDR2 socket will have 1207 pins up from 939 for that reason.

Btw Smithsfeld should arrive *with* Yonah which is a dual-cored Sonama. The EE version will have 4 logical CPUs as well. Intel is not pushing prescott, they are doing both Pressie and Dothan in tandem

is that why socket 940 has a registered RAM pin or is that something completely different? oh, never mind that's because their RAM controller is on-board. does anyone have the definite answer on this?

 

[xtknight]

Originally posted by: keysplayr2003

Originally posted by: Ackmed
Running 3-3-3 is idiotic, and I can think of only one reason why they would do that. To make the A64 slower.

Umm, nowhere in that article did anyone say they intentionally made the Athlon systems slower on purpose to level the playing field for the Pentium M platform. If I am wrong please show me. Its just starting to sound like reaching to me when someone grasps for little excuses as to why a processor does, or does not do so well.
So, how much of a difference is there from DDR333 @ 2.2.2 then DDR400 @ 3.3.3?
This is the question that will make or break the claim. Maybe.

Can we please make it our business to damper the flames (if any) in this thread.
So, lets take care of business folks. Thanks.

hmm...well I did some experimenting with memory timings a little while ago. here are my findings. this is with a pentium4, so i'm not sure if you can expect more or less difference on an Athlon 64. but the difference was very small, on my system at least. for isolation, the tests were run at the same DDR speed though.

http://forums.anandtech.com/messageview...hreadid=1480555&enterthread=y&arctab=y

Performance increase per -0.5 CAS : 0.50%
Performance increase per -1 RAS2CAS : 1.90%
Performance increase per -1 RAS PRECHARGE : 1.19%
Performance increase per -1 TRAS : 0.45%

so I can conclude that the DDR400 running at a high latency will definitely beat the DDR333, because the latency performance delta is neglibible. the total increase from 3-3-3 to 2-2-2 is 3.59% (on MY system). this won't go for every system but I wouldn't expect to see too much variance among Athlon 64 configurations. maybe Zebo can shed some light on this...

in other words, my data and common sense show their testing methods to be questionable. I'd think that DDR333 to DDR400 would yield more performance increase than 3.59%, but I could be wrong.

if they haven't scientifically tested differences in timing settings then they shouldn't just assume it will be equal. that's unprofessional. I'm not saying they didn't test it but it just doesn't seem like they did if what I stated above is correct.

 

[classy]

Originally posted by: keysplayr2003

Originally posted by: Ackmed
Running 3-3-3 is idiotic, and I can think of only one reason why they would do that. To make the A64 slower.

Umm, nowhere in that article did anyone say they intentionally made the Athlon systems slower on purpose to level the playing field for the Pentium M platform. If I am wrong please show me. Its just starting to sound like reaching to me when someone grasps for little excuses as to why a processor does, or does not do so well.
So, how much of a difference is there from DDR333 @ 2.2.2 then DDR400 @ 3.3.3?
This is the question that will make or break the claim. Maybe.

Can we please make it our business to damper the flames (if any) in this thread.
So, lets take care of business folks. Thanks.

In the original article with a different setup they ran the A64 at 333 but the Dothan at 2.5,3,3. Look here for setup.
Here are a couple of the game benchmarks, here. Look at the Doom 3 and Farcry numbers, especially Far Cry. The Dothan looks like it owns, huh?

But some peeps saw through the misleading of the first article so they rerun the benchmarks. Here is a quote from them

Others lashed out at our system testing configurations, claiming one side was biased and had advantages over the others. Some thought it was unfair that we used DDR-333 for Pentium-M testing, giving an advantage to AMD (when in fact, this is a limitation of the chipset, not our specific memory choice). Others thought that we should have used higher performance CAS 2 DDR memory for the Athlon64 platform, as the on-die memory controller of this chip really thrives with fast DDR, and that we were giving an advantage to Intel by not testing with faster memory.

The whole page of excuses is here.

Same setup just different video card.
Revisited setup here.

And lo and behold the same Dothan with same oc gets slapped, it does well but still kissing the rearend of the A64.

Doom3
Far Cry

Now I guess that proves their original benchmarks were either bogus, an outright lie, or a simple slip :roll:.

Now we move to their latest round of benchmarks on the 478 adapter and what do they do? Have the same garbage setup like they did in their original article. And the Nforce boards really churn with tight memory timings. Its a joke what they have put out. What's amazing is, why not OC the A64 as well. But they ain't gonna do that. I also find it funny how they get this 2.7 oc but no one else got that high. The whole article is fishy and phony IMO and I have been a fan of them as well. Maybe they are having problems selling their $925 Dothan CPU kit. :Q I still like them a lot, but it is clear they are misleading folks either through human error or on purpose. I'll let you decide.

 

[Duvie]

P4s and A64 are different and react totally different...The P4 is starved for bandwidth and I usually got no real difference with cas timings but with raw speed...the A64 has an overabundance of bandwidth it doesn't use now but likes the lower latency ram as per its controller....Cant compare the two....


To Dapunisher and PM....i was speaking based on HT as we know it today. If Intel is working on something different that is great. Obviously the HT was designed more as a crutch for the long idle waits and the latency penalties due to branch misprediction before in relation to the long pipeline of the northwood and subsequent prescott. A small stage cpu in the future will not be able to be based on this route and will be a drastically different in its implementation.....So basically we are not talking about the same thing...



Anybody who runs 400ddr on an A64 setup and runs cas 3-3-3 is either 2 things...1) a complete idiot of the architectures he is testing...or...2) trying to skew or spin the results...PERIOD...

Even the cheapest ram around can do cas 2.5 now with the DDR400 standard...No excuse....

Hey this is very common for reviews with thier predisposed biases to someone run the other ppls systems at less then ideal situations....IE run on via solutions not the more powerful NF3 or 4, run terrible cas timings, or manipulate bios settings....


I haven't read the full artcile but this would not be the first time in the last few months someone has done this.....

 

[imported_michaelpatrick33]

Originally posted by: keysplayr2003

Originally posted by: michaelpatrick33

Originally posted by: keysplayr2003

Originally posted by: classy

Originally posted by: Frown66
New article on Pentium M on the ASUS adapter. They got it to 2.7ghz. Its a great gaming cpu:

http://www.gamepc.com/labs/view_content.asp?id=770ct479&page=8

I'm just curious what this cpu (overclocking-wise) can do with a better cooling solution. That stock HSF for it looks rather wimpy.

You gotta be careful with Gamepc because they can mislead people. They run all the A64 memory at cas 333 because they say its because its not fair to the Dothan because it can only utilize ddr on a slower fsb. Yeah OK :roll: Thats why those marks look so impressive. Anand was running @2.56 and at that speed only "equaled" the A64 in a few benchmarks. So I question the validity of the marks of Gamepc. And most of all the other reviews have have shown similar results to Anandtech. Gamepc is the only odd ball with huge overclocks.

3-3-3 on the AMD 64. Who are they trying to kid? No one runs that high latency crap on the AMD 64. Additionally, I would be curious whether they have 1t or 2t command timings on the AMD 64. They don't state it, and if they set a 2t rate than the AMD would be 10-12% slower.

If AMD releases their dual core at 2.4 it will destroy Intel's 3.2 dual core as far as performance and thermal viability (I believe the dual core AMD's are 95w or 110w maxiumum wherease the Intel is 135w and of course AMD's is maximum rate [i don't know the technical terms, lol] while Intel's is an average rate). A FX-53/4000+ versus a Prescott 3.2 in single cpu performance is not even close and AMD's dual core actually sounds like truer "dual core" implementation than Intel's.

A dual core 2.4 AMD64 versus a dual core Dothan based 2.4 would be fun though!

Well, I see a lot of numbers and not much else. If your going to discredit something, make it undeniable. Get some facts together to disprove this article if you can.
Then we can safely say, "The Athlon Platform was crippled" which I doubt very much.

Uhh, where am I actively discrediting the article. I am not saying the article is wrong, only that the use of 3-3-3 DDR 400 memory on the AMD64 is a joke. I did not say the results were far wrong. Additionally, it is very easy to change the memory command rate from 1t to 2t and cause a dramatic drop in performance with the AMD64's. They did not state whether they set a 1t or 2t timing on their AMD 64 setup. Even experienced testers sometimes will forget at first to set their command rate to 1t. I am pointing out a missing variable in their methodology that in my opinion should be stated so as to not create a potential error in their tests and suspicion about their potential bias. I did not say their results were wrong! Just ... curious and with some interesting hardware timings and omissions.

Where are all my numbers looking to create an idea that I am stating their conclusion are false? I am pointing out that NO ONE using 3-3-3 DDR 400 (and it is a joke) and that the omission of command rate timings should not have been done, but their conclusions and benchmarks may very well be accurate within 1% or as far off as 15%. I don't know.

 

[imported_michaelpatrick33]

Originally posted by: DAPUNISHER
Windows XP professional supports 2 physical processors or 2 physical+2 logical which I think it treats as 4 logical processors braddah.

Yes, I believe MS switched their licensing from logical cpus to sockets. So the Extreme Edition Intel dual core with 4 logical processors, 1 socket - 2 cores - dual hyperthreaded will be supported.

 

[Ackmed]

Originally posted by: keysplayr2003

Originally posted by: Ackmed
Running 3-3-3 is idiotic, and I can think of only one reason why they would do that. To make the A64 slower.

Umm, nowhere in that article did anyone say they intentionally made the Athlon systems slower on purpose to level the playing field for the Pentium M platform. If I am wrong please show me. Its just starting to sound like reaching to me when someone grasps for little excuses as to why a processor does, or does not do so well.
So, how much of a difference is there from DDR333 @ 2.2.2 then DDR400 @ 3.3.3?
This is the question that will make or break the claim. Maybe.

Can we please make it our business to damper the flames (if any) in this thread.
So, lets take care of business folks. Thanks.

Have you ever seen anyone run a A64 system with 3-3-3? No. Any other reviewers use that setting? No. It is on purpose, there is no other reason to do so.

 

[RaynorWolfcastle]

Originally posted by: pm

Originally posted by: clarkey01
To back Duive up he's saying that the smaller pipeline in Yonah and Presler will not benifit from HT. Yes G5 use a form of HT and there only 17 or so pipelines deep, an Athlon 64 doesnt require or would benifit greatly from HT.

The CPU that I work on, Montecito, - an upcoming member of the Itanium family - has a fairly short pipeline and implements MT on both of it's cores. We see a substantial boost in some applications - particularly transactional computing.

I'm sorry for going a little OT, but how do you feel Intel's commitment to the IA64 architecture is these days? I think it's really too bad that it's getting such negative attention since VLIW offers a type of parallelism that cannot be achieved from going to multiple cores. As a matter a fact, I would speculate that as the number of cores goes up VLIW becomes increasingly more efficient compared to x86 multicores, no?