AMD's X2 A Failure Over At Tom's

[Continuity27]

Originally posted by: fatty4ksu
Wow, the 840EE is whoopin ass.

How so?

 

[V00D00]

Originally posted by: Lonyo

Because one is DDRII and the other is DDR?

Because one is 400mhz and one is 667mhz. 267mhz is quite a difference for memory.

They're using PC3200 (400mhz) for the AMD vs. PC5300 (667mhz) for the Intel.

3200
5300

That's a 2,100 MB/sec theoretical difference. That's why they're flawed.

edit
If you're comparing processors, why would you use memory with competely different bandwidths, that would clearly skew the results.

 

[biostud]

Are there any info on what OS the run?

Just wondering if they have the AMD X2 patch for winXP SP2 installed.

 

[JetBlack69]

Originally posted by: Lonyo
The OPERATING SYSTEM handles what the CPU does, not the CPU.
If the OS doesn't send the encoding thread to the CPU, there is NOTHING the CPU can do about it.
The OS could have diverted time from the RAR archive to encoding, or from the game test, or the CD rip, and increased the amount of encoding done, but it didn't.
This is not the fault of the CPU, it is the fault of the system scheduler.
The CPU multitasks properly, the test does not, and the OS does not.
You would not rip the same CD over and over, you would not RAR and archive over and over, you would probably change focus to give the encoder higher priority.

The CPU is not at fault for OS issues.

:thumbsup: I'm glad some people here know what they are talking about.

 

[imported_g33k]

Originally posted by: V00D00

Originally posted by: Lonyo

Because one is DDRII and the other is DDR?

Because one is 400mhz and one is 667mhz. 267mhz is quite a difference for memory.

They're using PC3200 (400mhz) for the AMD vs. PC5300 (667mhz) for the Intel.

3200
5300

That's a 2,100 MB/sec theoretical difference. That's why they're flawed.

edit
If you're comparing processors, why would you use memory with competely different bandwidths, that would clearly skew the results.

DDR2 is what normally runs on the intel system, DDR is what normally runs on the amd system. How does this skew the results?

 

[AnonymouseUser]

Originally posted by: g33k

Originally posted by: V00D00

Originally posted by: Lonyo

Because one is DDRII and the other is DDR?

Because one is 400mhz and one is 667mhz. 267mhz is quite a difference for memory.

They're using PC3200 (400mhz) for the AMD vs. PC5300 (667mhz) for the Intel.

3200
5300

That's a 2,100 MB/sec theoretical difference. That's why they're flawed.

edit
If you're comparing processors, why would you use memory with competely different bandwidths, that would clearly skew the results.

DDR2 is what normally runs on the intel system, DDR is what normally runs on the amd system. How does this skew the results?

Exactly. Using the highest-rating recommended components for each system in a direct comparison is completely acceptable.

 

[Continuity27]

This test has become very peculiar... :Q

 

[justly]

I just noticed the Intel sysytem DivX scores just made a big jump. So to correct my last post it would seem that the Intel systems ability to activily work on more threads than the AMD system at one time (as it should with 4 virtual cores) does benifit Intels DivX score.
Using this senerio (and updated scores) it would now appear that Intel is processing DivX faster, (considering run time) about equal or slightly behind AMD on the Lame test, and way behind AMD in WinRAR, Farcry and stability.

I would have to agree with Lonyo on this point.
"Basically the test is 100% useless for anything except showing system stability"
Since this senerio is quite unrealistic.

 

[Lonyo]

Originally posted by: AnonymouseUser

Originally posted by: g33k

Originally posted by: V00D00

Originally posted by: Lonyo

Because one is DDRII and the other is DDR?

Because one is 400mhz and one is 667mhz. 267mhz is quite a difference for memory.

They're using PC3200 (400mhz) for the AMD vs. PC5300 (667mhz) for the Intel.

3200
5300

That's a 2,100 MB/sec theoretical difference. That's why they're flawed.

edit
If you're comparing processors, why would you use memory with competely different bandwidths, that would clearly skew the results.

DDR2 is what normally runs on the intel system, DDR is what normally runs on the amd system. How does this skew the results?

Exactly. Using the highest-rating recommended components for each system in a direct comparison is completely acceptable.

That's what I said, but v00d00 decided to only quote one part of my post and ignore the rest.

 

[mechBgon]

Originally posted by: ProviaFan
By the way, what if I set up a benchmark with 5 threads on the P4.

Exactly. The 840EE is the only P4 with both HT and dual-core, thus four virtual CPUs, so picking four CPU-intensive tasks makes ya go

~ hmmm, it's almost as if they are trying to optomize the test to suit the Intel processor...

Let's see them do that test with a Pentium D (dual-core, no HT). If the problem is really with the CPU architecture and not the OS thread scheduling, that should reveal it. But I have a feeling it would reveal something else instead :evil:

We all know AMD CPU's are poor at floating point operations

We do? Could've fooled me, AMD = strong FPU ever since the K7 came out.

 

[RanDum72]

We do? Could've fooled me, AMD = strong FPU ever since the K7 came out.

Yep. Check here. The Opteron has the strongest overall FPU performance.

 

[boran]

this scheduling issue makes the whole thing useless tho, AMD wins in 3 of the 4 benches, but how much more is that latter bench worth, in CPU what if the scheduler got it right ? lots of questions. and as mentioned before, intel simulates 4 processors which makes the OS much more multithread-capable. and I think the video encoding thread is set to a lower priority than the other threads. imho the whole test is rendered worthless because of this.

 

[EightySix Four]

Man the pentium D EE is a complete failure... Look at how few winrar's it's done, shows how weak it's multitasking performance is... It's USELESS!

 

[Pabster]

Originally posted by: crazySOB297
Man the pentium D is a complete failure... Look at how few winrar's it's done, shows how weak it's multitasking performance is... It's USELESS!

Yep, D is definitely a total flop.

LOL.

edit: the LOL signifies how stupid the poster is, not to realize the CPU being pitted against the X2 4800+ is an EE 840, not the D. Heh. Fanboys are so easy to spot.

 

[Pabster]

Originally posted by: boran
this scheduling issue makes the whole thing useless tho, AMD wins in 3 of the 4 benches, but how much more is that latter bench worth, in CPU what if the scheduler got it right ? lots of questions. and as mentioned before, intel simulates 4 processors which makes the OS much more multithread-capable. and I think the video encoding thread is set to a lower priority than the other threads. imho the whole test is rendered worthless because of this.

The thing is, think about it. Regardless of how many processes or threads.

We have WinRAR, LAME, FarCry, and DiVX running. All simultaneously.

Obviously the X2 is devoting 99% to the first 3 tasks and nearly nothing to the last.

The EE, however, appears to be dividing fairly porportionally across each.

So let us figure that if the X2 DID devote an equal portion to that last thread (DiVX) it would not have any (or very little) lead over the EE in the other 3 tasks.

 

[Howard]

I friggin hate stumbling into threads that lower my IQ. I want my fvcking IQ back!

 

[deveraux]

Originally posted by: Pabster

Originally posted by: boran
this scheduling issue makes the whole thing useless tho, AMD wins in 3 of the 4 benches, but how much more is that latter bench worth, in CPU what if the scheduler got it right ? lots of questions. and as mentioned before, intel simulates 4 processors which makes the OS much more multithread-capable. and I think the video encoding thread is set to a lower priority than the other threads. imho the whole test is rendered worthless because of this.

The thing is, think about it. Regardless of how many processes or threads.

We have WinRAR, LAME, FarCry, and DiVX running. All simultaneously.

Obviously the X2 is devoting 99% to the first 3 tasks and nearly nothing to the last.

The EE, however, appears to be dividing fairly porportionally across each.

So let us figure that if the X2 DID devote an equal portion to that last thread (DiVX) it would not have any (or very little) lead over the EE in the other 3 tasks.

The thing is though, neither EE nor X2 is deciding which task to perform. The OS is. If the OS doesn't want to send the instruction to the CPU to perform the task, the CPU doesn't even get a chance to do it. I can't vouch for whether the X2 will perform the task better with the thread priority set to Normal or equal to the others since I don't have one and have never benchmarked it. However, you are speculating that the X2 won't perform better than the EE given a fair chance.

If say, Intel released a single core EE chip with HT that the OS detects as 4 logical CPUs instead of 1, a similar result will occur since with the Intel chip. No matter what, the DivX encoding will still receive some CPU time just because the OS will schedule it some CPU time. Does that make the single core EE faster than the X2? Not in my book since the OS isn't giving the X2 a fighting chance.

I say, just set the bloody thread priority for the DivX encoding to the same level as the other tasks and let's all see the true result of which CPU is better.

My $0.02.

 

[Pabster]

Originally posted by: deveraux
I say, just set the bloody thread priority for the DivX encoding to the same level as the other tasks and let's all see the true result of which CPU is better.

Fair enough, I say go for it.

 

[Googer]

Originally posted by: porkster

Originally posted by: justly
Right now I would have to say the DivX number are useless, hopefully toms will clairify what is going on with them...

I have to agree, but if the divx encoding is having problems on the AMD X2 then you can obviously admit the other values are under rated for the Intel 840EE.

It will be interesting to hear what Tom has to say about the faults on the AMD X2, as they clearly need fixing before the AMD X2 goes to market.

Others should warn their friends of the design fault of the AMD X2.

.

Who says the CPU is to blame? It could be a host of other factors that is preventing it from operating correctly, Prehaps Microsoft is to blame and needs to issue a critical update for X2 users. It could be differances in the two motherboards. Who knows, dont be quick to blame the obvious.

 

[Googer]

Originally posted by: otispunkmeyer
thos cant be live results either......

munich isnt behind england in time, its ahead. its 11am here now, and their little graph thing says 2-3am so wth?

They update them selves every minute.

The server is located in a differant timezone dummy. It is reading the correct time for me.

 

[Googer]

Originally posted by: carlosd
THG article is crap, it is plagued with major insconsistencies. How can yo read than kind of BS?

Then please explain how it is full of inconsistancies? I however see little or none. Otherwise you are full of BS for posting Cr@p like this.

 

[Googer]

Originally posted by: V00D00
Did anyone notice how the Intel System got DDRII 667mhz, while the AMD system got 400mhz.

Ok, Well there you go. I assume WinRAR is about 99% memory bandwidth dependant.

Right there the benchmarks are flawed. Why would those retards use such different memory????

It's not toms fault, Intel's newest 945 Chipset only Supports DDR2. Now you tell me where I can Find an AMD X2 processor and motherboard that supports DDR2 and I will buy it and send it directly to tom. Even better I will hand deliver it.

In otherwords the Intel system cannot work with out DDR2.

 

[Dough1397]

why is the intel machine getting 4 programs and windows scheduler sees 4 cores, where as amd x2 gets 4 programs and sees two cores.... that a bit sad.... mayeb teh x2 would be better at doign 2 things at once and finishing quickl and doign the otehr 2 and then be done near the same times as the intel system took to do 4 tasks starting all at the same time!?

 

[Googer]

Originally posted by: Dough1397
why is the intel machine getting 4 programs and windows scheduler sees 4 cores, where as amd x2 gets 4 programs and sees two cores.... that a bit sad.... mayeb teh x2 would be better at doign 2 things at once and finishing quickl and doign the otehr 2 and then be done near the same times as the intel system took to do 4 tasks starting all at the same time!?

I might agree with you, It is unbalanced in that sence. the test should be run with Hyperthreading turned off, that way the os sees 2 processors on both systems. This is where the real inconsistancy lies.

 

[EightySix Four]

Originally posted by: Pabster

Originally posted by: crazySOB297
Man the pentium D is a complete failure... Look at how few winrar's it's done, shows how weak it's multitasking performance is... It's USELESS!

Yep, D is definitely a total flop.

LOL.

edit: the LOL signifies how stupid the poster is, not to realize the CPU being pitted against the X2 4800+ is an EE 840, not the D. Heh. Fanboys are so easy to spot.

Hehe actually it's a dual core EE with 2 cores plus hyperthreading... How Skewed is that.