Socket 939 Sempron found........

[TGS]

Intel:

DivX: 0.6%
Farcry: 46.2%
Lame Encoder: 21.0%
WinRAR: 46.2%
Load CPU1: 100.0%
Load CPU2: 100.0%
Load CPU3: Not Available
Load CPU4: Not Available


Overall 114%. Nice!


AMD:

DivX: 1.3%
Farcry: 38.2%
Lame Encoder: 0.0%
WinRAR: 29.5%
Load CPU1: 98.6%
Load CPU2: 99.6%


Overall 69%. Boo.



Updated

DivX: 0.0%
Farcry: 25.1%
Lame Encoder: 39.5%
WinRAR: 38.0%
Load CPU1: 100.0%
Load CPU2: 99.1%
Load CPU3: Not Available
Load CPU4: Not Available

Overall: 102.6% WOOT



DivX: 0.5%
Farcry: 44.2%
Lame Encoder: 7.2%
WinRAR: 24.7%
Load CPU1: 98.9%
Load CPU2: 100.0%

Overall: 76.6% Hiss

Intel is obviously better, it can be at more than 100% utilization, while AMD is being lazy... Slacker.

 

[boran]

those percentage numbers say (as most of the data) exactly nothing. run an app and see the processor utilisation, run two and look again, it fluctuates all over the place. the only interesting item would be processor time (ie how many seconds would the processor have spent on the app at 100% CPU) my appreciation of THG sinks to new lows, I really liked the site untill this kinda junk. with their blaming the AMD memory controller, not that I say it cant be that, but it was way too soon to say something like that and guys working in a "lab" should at least apply some form of scientific methodology, which is dont say anything as true untill you know for 99% sure it is.

And the memorycontroller thing was pure speculation. and seeing now how far the divX thread progresses it was also far from the truth.

 

[Markfw]

It is beating the pants of the Intel, so if you think its a slacker, its a slacker kicking Intels a$$. I don't believe those percentage numbers.

 

[justly]

Originally posted by: boran
those percentage numbers say (as most of the data) exactly nothing. run an app and see the processor utilisation, run two and look again, it fluctuates all over the place. the only interesting item would be processor time (ie how many seconds would the processor have spent on the app at 100% CPU) my appreciation of THG sinks to new lows, I really liked the site untill this kinda junk. with their blaming the AMD memory controller, not that I say it cant be that, but it was way too soon to say something like that and guys working in a "lab" should at least apply some form of scientific methodology, which is dont say anything as true untill you know for 99% sure it is.

And the memorycontroller thing was pure speculation. and seeing now how far the divX thread progresses it was also far from the truth.

I totally agree.

Also, it would seem Duvie (and many others) where right about the windows scheduler all along . Was there ever a doubt?

 

[Jeff7181]

Originally posted by: porkster

Originally posted by: Jeff7181I think Intel is doing a huge disservice to themselves and their customers by disabling HT in the Pentium D. As you can see, from the results so far, a dual core 3.2 GHz Pentium D (which is basically what that 840EE has become with HT off) is handily being thumped by the 4800+.

HT with dual cores is the same exact story as it was with single cores. It allows for heavier multi-tasking as it's capable of running 4 threads simultaneously vs. the X2 4800+'s 2 threads simultaneously.

*EDIT* It's the exactly problem described by Anand in his review. Windows sucks at preemptive multitasking.

Totally agree.

The presscott cpu's are such great overclockers so I don't think Intel wants customers buying the low end and getting top end performance, a reason maybe why they disable features like HT in the low end of their product range.

We are still yet to know what priority level the dvd-to-divx thread is, so you can't blame the OS yet.

.

Windows gives the foreground application priority. In THG's tests, doesn't he say the game is the foreground application? That would mean the game gets the highest priority. I assume the rest of the tasks are set to normal priority and for some reason Windows doesn't give each task an equal amount of CPU time.

 

[Duvie]

Originally posted by: porkster

Originally posted by: AkumaXedit: and now the AMD system's changing load balance...

DivX: 0.9%

Both are below 1.0, so it looks like the problem is the priority level and the way the divx encoding is processing. Early data would say they will both be neck and neck on the task.

The test now is to force the dvd-divx thread to a high priority and turn on HT on the Intel and see which CPU performs the best. The proper stress test since they ironed out most of the issues.

.

Ding Ding Ding....exactly what I was saying all along....

 

[Duvie]

Originally posted by: TGS
Intel:

DivX: 0.6%
Farcry: 46.2%
Lame Encoder: 21.0%
WinRAR: 46.2%
Load CPU1: 100.0%
Load CPU2: 100.0%
Load CPU3: Not Available
Load CPU4: Not Available


Overall 114%. Nice!


AMD:

DivX: 1.3%
Farcry: 38.2%
Lame Encoder: 0.0%
WinRAR: 29.5%
Load CPU1: 98.6%
Load CPU2: 99.6%


Overall 69%. Boo.



Updated

DivX: 0.0%
Farcry: 25.1%
Lame Encoder: 39.5%
WinRAR: 38.0%
Load CPU1: 100.0%
Load CPU2: 99.1%
Load CPU3: Not Available
Load CPU4: Not Available

Overall: 102.6% WOOT



DivX: 0.5%
Farcry: 44.2%
Lame Encoder: 7.2%
WinRAR: 24.7%
Load CPU1: 98.9%
Load CPU2: 100.0%

Overall: 76.6% Hiss

Intel is obviously better, it can be at more than 100% utilization, while AMD is being lazy... Slacker.

Dont be a moron!! Obviously the chart has something not working cause lame encoder is obviously getting work done equivalent to Intel right now...

I dont care what the percentage load says but that the cpu (1&2) stays loaded at 100% (which there are you genius) and what the totals are....

NOw that stability has been hammered out for both this really isn't a stress test but a performance test now....


Also why does Toms include runs if they want to use fps as the measuring tool?? What a bunch of idiots...FOS has seen in the first two rounds of testing fluctuates wildly for Intel from 27-37 whereas AMD styed pretty consistent in the 35-36 range....The chart tells a perosn nothing but number of runs tells a person who was able to hold the higher fps on average longer...

 

[boran]

I think if the divx thread is actually set to lower (we can see in a few hours) then MS has to work to do with winXP, because their thread sheduler cant handle 2 x 2 virtual cores. you see the behaviour as it is now is exactly how it would behave is divX was set to a lower thread priority. if anyone remembers the previous results, I'd think you'll find intels current performance in the three main areas to be better than it was before.

if that's the case I think we can conclude that HT is not supported on dualcores by winXP atm and it lets other threads steal your performance away.

I repeat before th intel zealots curse me to hell: the issue lies with winXP.

edit and I really think TGS was being sarcastic...

edit2: does anyone know how much songs that CD counts ? I'd like to watch the thing and test a theory but I need to know the number of songs.

edit3: enyone noticed the power usage is lower. I wonder if that's because of no HT.

 

[Duvie]

Actaully CD encoding is worse for INtel now...Intel upon the stable reboot usually per 25 encodings would have 3 more then AMD...Winrar looks near the same interms of percentage...Gaming is hard to tell right now cause INtel has so much fluctuation over a course of 24 hours...This one we will have to look at later with some time behind it...

Divx is obviously doing more like AMD now...can you say slugs??? I imagine it may be 10 minutes per 3-4 hours....

 

[boran]

I'm surprised actually the divX gets done that much, that means the other three processes are doing nothing long enough to get 10 mins of divX encoded.

 

[Markfw]

Again, AMD tied on two, winning on two, winrar by quite a bit !

15 each on LAME

93 Intel, 122 AMD on winrar,

68 Intel, 73 AMD (and one higher fps) farcry

10 each dixv.

 

[SlitheryDee]

Originally posted by: Duvie

Originally posted by: porkster

Originally posted by: AkumaXedit: and now the AMD system's changing load balance...

DivX: 0.9%

Both are below 1.0, so it looks like the problem is the priority level and the way the divx encoding is processing. Early data would say they will both be neck and neck on the task.

The test now is to force the dvd-divx thread to a high priority and turn on HT on the Intel and see which CPU performs the best. The proper stress test since they ironed out most of the issues.

.

Ding Ding Ding....exactly what I was saying all along....

I think tom's probably knew as well as we did that the problem was with the windows scheduler.

All that BS in the 6th update was merely to stir us up and keep us talking about it.

Edit: But then again I could be giving them too much credit...

 

[Duvie]

Too bad they didn't leave the temperature gauge on so we can see how much HT effected the PD's temp.....I can only assume it still wasn't great and so Toms is trying to shield it...

 

[Duvie]

Originally posted by: Markfw900
Again, AMD tied on two, winning on two, winrar by quite a bit !

15 each on LAME

93 Intel, 122 AMD on winrar,

68 Intel, 73 AMD (and one higher fps) farcry

10 each dixv.

actually every time the iNtel CD encoding will swicth up one before the AMd does....I think Intel is ahead by a slight margin...2-5% range...we will see once enough time can get the numbers up a bit....Divx is the same way...

 

[justly]

Originally posted by: boran

edit3: enyone noticed the power usage is lower. I wonder if that's because of no HT.

I think that is a safe bet.

 

[porkster]

Originally posted by: boran
I think if the divx thread is actually set to lower (we can see in a few hours) then MS has to work to do with winXP...

Yes it seems like Windows isn't handling the threads very well when then are less cpu pipes than application threads for multiprocesser systems.

.

 

[mountcarlmore]

wow porkster, you have seen the light. on that note, i agree, the intel is better at mutli-tasking cause of hyperthreading. however, this case of mutli-tasking is unrealistic. when i mutli-task, i would most likely be donig only two things, like playing a game and encoding a dvd, for which amd will suit me better, cause it games better, and as you might like to know, amd is actually very good at divx, beating intel on most of the reviews, of course depending on which program is being run. if you need to run 4 simultaneous taks, you have no business being on a desktop, thats what a 4 way server is for, cause i doubt any home user is gonna multi task like that.

 

[Pabster]

Wait, HT is OFF now. And the 840 EE is performing marvelously. Perhaps this lends creedence to the observations of various people that the EE 840 actually is HURT by HT in many scenarios rather than helped?

 

[SlitheryDee]

Look at this

If you haven't read it this is from the tom's hardware "damage control" page that porkster linked earlier.

Quote:

Particularly when you're comparing two interchangeable components, there are two basic principles to which you need to adhere. The first is to never compare apples to oranges. The basic components used in the test should never be changed in the middle of the test. If, for example, you're comparing several types of graphics cards, then the motherboard, CPU, RAM, hard drive, etc., shouldn't vary. Likewise, it's taboo to change the settings in the middle of the test. Second, you should try to have comparative values for similar, well-known products at hand. The reliability and overall value of your benchmarks have a lot to do with how much you follow these principles.

An entirely different concept of benchmarking involves comparing two completely different computers. In this case, it is okay to compare apples to oranges."

linky

Notice the sentence in bold.

Edit: Looks like they're excusing their own mistakes to me...

 

[Zebo]

I hope this gets enough attention to change the quality of their site in future! Terrible just terrible bias: Intel crashed over twelve times if you include the two Intel motherboard crashes and the Epox board which they said would'nt boot.

Ok you guys need to view these with internet explorer since they are mht archive files. I had some help from Carlos archiving these:

Ok first off the fan myth. The fan they used was correct at first, notice high rpms, and big bore used

First off we have the inital gigabyte board with already one crash

Next Toms switches boards to Asus NF4 3 reboots later as shown

Tom's can't have that!!! So he resets the boot counter back to 1!!!

Crashes 4 more times as shown here before finally upgrading to Asus 955x board. That's at least 7 crashes so far not including the two inoperable Intel boards and EPOX baord!!!

Meanwhile, as shown here AMD is rock solid 97 hours straight!!! (actually more like 102 hours but I don't have an archive of that)

Toms does'nt like Intel 7-9 crashes so he resets counter yet again to Zero for the intel system and whipes out AMD rocksolidness restarting all testing

Restarts all testing, with Asus 955x, Intel Crashed 3x since with the new testing on Asus 955x due to toms putting on the wrong heat sink @ 1200 rpms

Count them up to 12 crashes for intel, Zero for AMD.

You think that's bad? I do.

Bottom line: Toms HW is clueless on getting a system running, that any HS student can do.. In light of this there is no way I trust thier benchmark programming or tabluation methods which are much more difficult than building a system, especially in light of the "no network" issues. I've stopped replying to the "other" thread in forum because really what's the use? there's like 10 other "good" sites out there Im busy reading instead...techreport, anandtech, xbit etc.

 

[MDme]

HT has always been a double edged sword. With HT the windows scheduler is able to put in 4 threads in the 2 real + 2 virtual cores of the P XE. But it does come at a price: the other threads suffer a bit. In the case of the THG tests, the winrar and farcry tests suffer. If you look at the live stats now, you can actually see that the X2 lead in winrar and farcry are less (vs HT enabled).

It would actually be nice if THG tried to put all threads in the same priority (albeit manually) and ran the test with HT enabled. I think that way we can see how HT behaves and impacts the system (good or bad). even better if they could run it without HT too so we can compare.

 

[Markfw]

Originally posted by: Pabster
Wait, HT is OFF now. And the 840 EE is performing marvelously. Perhaps this lends creedence to the observations of various people that the EE 840 actually is HURT by HT in many scenarios rather than helped?

Where do you get marvelous out of this ? It was an argueable tie, sawying to AMD before, now AMD is tied on two, ahead on the other two !!

Edit: the Intel in finally a little ahead in lame, so 2 AMD but a large margin, One Intel by a small margin, and one tie.

 

[Zebo]

Originally posted by: Markfw900

Originally posted by: Pabster
Wait, HT is OFF now. And the 840 EE is performing marvelously. Perhaps this lends creedence to the observations of various people that the EE 840 actually is HURT by HT in many scenarios rather than helped?

Where do you get marvelous out of this ? It was an argueable tie, sawying to AMD before, now AMD is tied on two, ahead on the other two !!

Edit: the Intel in finally a little ahead in lame, so 2 AMD but a large margin, One Intel by a small margin, and one tie.

You guys continuing to argue over Toms incompitance let alone trust anything he publishes?

I mean for christ sakes he's not even able to show temps or rpms properly anymore.
http://users.pandora.be/compas/stresstest/intel_bestanden/cam2.jpg

well it's true: http://www.tomshardware.com/stresstest/intel.html

Not to mention every other site on the net disagrees with his performance numbers..:roll:

 

[SlitheryDee]

Originally posted by: Markfw900

Originally posted by: Pabster
Wait, HT is OFF now. And the 840 EE is performing marvelously. Perhaps this lends creedence to the observations of various people that the EE 840 actually is HURT by HT in many scenarios rather than helped?

Where do you get marvelous out of this ? It was an argueable tie, sawying to AMD before, now AMD is tied on two, ahead on the other two !!

Edit: the Intel in finally a little ahead in lame, so 2 AMD but a large margin, One Intel by a small margin, and one tie.

The intel system was outpacing the x2 (ever so slightly) in lame before they disabled HT .

It seems that disabling HT doesn't significantly change the standings in the benches (except for divx encoding).

I must say that I am extremely curious about what would happen if they enabled HT and raised the priority of gordian knot for the amd system.

 

[Pabster]

Originally posted by: Markfw900
Where do you get marvelous out of this ? It was an argueable tie, sawying to AMD before, now AMD is tied on two, ahead on the other two !!

Calm down Dothan (oops, sorry, with all that excess !! I mistook your identity )

Perhaps the word "marvelous" was too strong. Let me say the Intel machine is performing better in 2 of the tasks than it did with HT enabled. Is that OK?

Edit: the Intel in finally a little ahead in lame, so 2 AMD but a large margin, One Intel by a small margin, and one tie.

Interesting how one's opinions can shape their perspective, isn't it?