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

Page 17 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

kitkat22

Golden Member
Feb 10, 2005
1,464
1,333
136
I was reading over the current update and found a semidecent explanation to everything except one line that caught my attention, "Both systems are neck-and-neck on the other applications - MP3 audio encoding and packing with Winrar." Really? From what I have seen and watched for the past several days the winrar test is taken by AMD by a large margin. There is a 28% difference between the two machines AMD sporting 2635 and Intel 1890. I fail to see neck-and-neck, can anyone else see it?

They did give AMD the Farcry lead by 4fps. Furthermore, the AMD machine has completed more loops. The CD test is literally neck-and-neck and we have already seen and can explain the issue with divx. I've lost faith in this stress test, but I'm hoping they will do better and gain some credibility with the "performance test" they will now be running.

(Sorry, I did my math wrong in terms of context.)
 

Keysplayr

Elite Member
Jan 16, 2003
21,219
54
91
Originally posted by: dug777
Originally posted by: GuitarDaddy
Although this online test has been a miserable failure up to this point, I think it has opened a lot of peoples eyes about the methods or lack of that THG uses during their review process. And the total lack of credibility in there motives and conclusions. I understand better now why some of Toms reviews seem so far off the mark and don't agree with other credible reviewers results. They have absolutely no clue about control procedures, and their tests are obviously designed with a particular result in mind. If they don't get the result they expected, they just change the test and start over:Q

to be fair to them they had to start over to keep the test fair ;)

Although the constant hardware failures haven't been their fault...i come out of this rather worried at the stability of the intel platform. I would love anand to do a similar test, to see if he can confirm these severe stability problems, it really can't be that hard to do it right can it?

Not hard. Not at all. There just has to be more than just one AMD rig and one Intel rig. I would feel better if there were at least 3 Identical Intel and 3 identical AMD rigs running side by side. 3 is not even enough for a stability test, there should be dozens. But what review site has the time or money to accomplish such an extensive test bed?

 

Pabster

Lifer
Apr 15, 2001
16,986
1
0
Originally posted by: keysplayr2003
Not hard. Not at all. There just has to be more than just one AMD rig and one Intel rig. I would feel better if there were at least 3 Identical Intel and 3 identical AMD rigs running side by side. 3 is not even enough for a stability test, there should be dozens. But what review site has the time or money to accomplish such an extensive test bed?

I'd imagine most, especially THG. They get most everything given to them. I'm sure if they WANTED 3 of each, it could be arranged.
 

The Pentium Guy

Diamond Member
Jan 15, 2005
4,327
1
0
Bah - Who cares about THG anyways, there's plenty of other places to get information. For now, the X2 is a clear winner across the board, with Intel maybe having an advantage in Video encoding as usual.
 

Pederv

Golden Member
May 13, 2000
1,903
0
0
Wow, what a surprise. My account is still active. Haven't been on in years.
Gotta take any test like this with a grain of salt. Would be nice if Anand would do a simular test, so we could see another point of view (A little better organized though).
 

porkster

Member
Mar 31, 2004
141
0
0
Originally posted by: The Pentium Guy
Bah - Who cares about THG anyways, there's plenty of other places to get information. For now, the X2 is a clear winner across the board, with Intel maybe having an advantage in Video encoding as usual.

The THG clearly showed the AMD X2 isn't the winner claimed in other site's solo thread tests.

Anyone that monitored the THG stress and multitasking bench test, would have noticed the Intel was hard to match when running correctly.

The overall stress test was a botch up, with THG being such lamers and not using new thermal pad, the wrong fan and the wrong fan connector on the motherboard. Also once they got the Intel system setup properly they didn't reset the score of the multitasking challenge making the whole event a waste of time.

The overall picture that comes out the THG test is that the AMD X2 is a poor multitasker and not much a match for the top end Intel desktop CPU's, but probably a good upgrade path for those already with a mobo that are AMD based.

The price of the CPU's and their over ability to do current applications, doesn't make them very relative for customer's needs. Even the rich would be stupid to waste the cash on these CPU's with out a real need for the grunt.

Lets hope some other sites can test both CPU's in the future with a similar real test that puts the strain on the systems.

.
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
The overall stress test was a botch up, with THG being such lamers and not using new thermal pad, the wrong fan and the wrong fan connector on the motherboard. Also once they got the Intel system setup properly they didn't reset the score of the multitasking challenge making the whole event a waste of time.

Please link that lie since I have not heard them mention this.....

They reset the test the first time (after the first 4-5 reboots) you fvcking retard...after the AMD had already had LIKE 3 PLUS DAYS of uptime....


How much does Intel pay you to sit at one of their PCs and spread FUD and X2 negativity??? seems like a rash of new trolls and I put it beyond kids off for the summer....

Facts....

Intel dual core boards are extremely unstable (NF SLI)

INtel bundled wrong HSF with the chip...morons

INtels performance is hard to get a read on cause it changes so much....first 3 hours it does 36fps in Farc cry and then the next day or so it only does 27 and now back to 35-36...I find that suspicious....CD encoding takes off then once it has a decent lead it slows down and speeds up the Far Cry....I would love to see if Toms is in their manipulating the data...Notice how the divx was running at 250:10 ratio in encoding time then it slowed all the way down to like 110:10...How convenient as it allowed cpu cycles to free up to help CD encoding and far Cry....

The first test had over 12 hours of stability in the beginning and the results now differ from that significantly and well beyond common platform change...AMD won 3 of them and held continuous leads and growth..temps were similar then as well with the INtel cpu....The 2nd round started differed tremendously from the 1st round...series of crashes later the board starts well again but then tails off in Far Cry as I mentioned...Does that for like1-1/2 days and now speeds up in all the right areas to close gap on test....

Yeah I am calling Toms a liar, fake, incompetent...how can anyone trust any numbers from him???? he is reading this thread....I know cause as soon as we start biatching about no network activity and potential falsities it magically starts working....

You listening tom?? Quit manipulating the data...the randomness of speed up and slow downs and all in the convenient areas of Intel are far to obvious.....You have no Cred cause your an idiot in most ppls book.....


I applore someone with cred like anandtech or TechRepot to do a test and I will put my stock in it...




 

Sqube

Diamond Member
Dec 23, 2004
3,078
1
0
Porkster... let me put it this way. Assuming that everything you say is true (a laughable assumption in and of itself, as you're clearly doing this to get attention and start a flame war), I would still pick the X2. Want to know why?

What the bloody fvck would I do with a machine that's unstable??? I mean seriously. I'd rather take a Corvette that runs over a Porsche that spends half its time in the shop. If I have a car, I want to drive it. If I have a computer, I want to use it, not play "Guess what made the machine crash this time."
 

Pederv

Golden Member
May 13, 2000
1,903
0
0
I've looked through this thread and I don't think anybody brought up the point that the Divx graph should be showing how many times the DVD has been compressed instead of how many minutes are spent on the application. It's possible that even though the Intel system is spending more time on Divx, it's not getting anything done because of the disk accesses need for Winrar and Lame.

Just a thought.
 

michal1980

Diamond Member
Mar 7, 2003
8,019
43
91
no one should listen to porkster, his a noob, just out for reaction. his doing the same crap at thg forums, just ignore him and eventually he might get bored and leave.
 

Pabster

Lifer
Apr 15, 2001
16,986
1
0
Obviously Tom's people are total idiots. Not being able to properly cool the P4 EE can't simply be attributed to Intel. Did he really think a 2000rpm fan would be sufficient (with a STOCK heatsink)? I mean, come on. Maybe he needs to do another mock-up review of the latest HS/F assemblies and then put a decent one on the P4 EE before he does this ridiculous "stress test".

As for the AMD fanboys who continue to mention "all those boards the P4 went through" keep in mind that 2 of them were SLI boards, and nVidia is having trouble with the Intel SLI platform. That's already been shown. The Epox board had faulty PCB components. The EE simply overdrew the voltage regulators and blew the board. You can see that pretty clearly.

So chalk this up to THG's inability to properly configure a machine. Many, many other sites have tested the P4 EE without 1/100 of the issues. Perhaps another site will do their own "stress test" with a bit more quality control.
 

justly

Banned
Jul 25, 2003
493
0
0
Originally posted by: Duvie

...INtels performance is hard to get a read on cause it changes so much....first 3 hours it does 36fps in Farc cry and then the next day or so it only does 27 and now back to 35-36...I find that suspicious....CD encoding takes off then once it has a decent lead it slows down and speeds up the Far Cry....I would love to see if Toms is in their manipulating the data...Notice how the divx was running at 250:10 ratio in encoding time then it slowed all the way down to like 110:10...How convenient as it allowed cpu cycles to free up to help CD encoding and far Cry....

Duvie, you have more experiance with these types of apps than me, so I was wondering if you think this could have anything to do with how HT shares cache?
Could it be, as the treads get issued 2 heavily cache dependant apps could be assigned to different virtual cores that share the same cashe thus slowing them down (slightly) and allowing the other two less cache dependant apps speed up while shareing the other cores cache using its two virtual cores, then swithing around and causing the slower apps to speed up and the faster apps to slow down.

Other than your conspiracy theory about Toms this is the only explaination I can think of.
 
S

SlitheryDee

Originally posted by: Pederv
I've looked through this thread and I don't think anybody brought up the point that the Divx graph should be showing how many times the DVD has been compressed instead of how many minutes are spent on the application. It's possible that even though the Intel system is spending more time on Divx, it's not getting anything done because of the disk accesses need for Winrar and Lame.

Just a thought.


I don't think it's minutes spent on the application. It's minutes of DVD video that have been encoded.

That means the amd system has compressed slightly over one DVD (assuming ~120 min movie length) while the intel system has compressed about 15.
 

kitkat22

Golden Member
Feb 10, 2005
1,464
1,333
136
Should I? Yes, I feel in the mood.

Porkster, I would like to point out a logical fallacy in your amazing deductive powers. You have just concluded, and rightfully so, that THG didn't do a great job on the test; lack of control, setup issues, etc, etc. Now in the same thread you start talking about the numbers and pointing out that Intel is higher/better and the machine to buy. Didn't you just come to the conclusion that the numbers may very well be wrong in and of themselves? If this were done in the scientific community the whole process and results would have to be thrown out because of botched experimentation. In other words, the numbers can't be trusted. My current stance is for THG to write down and perform an appropriately followed methodology of experimentation, or to have another site perform a test. In conclusion, your high and mighty opinion has no leg to stand on. As is said among the statistics world, "you can torture numbers till they tell you what you want to hear,? and apparently you have tortured them long enough.
 
S

SlitheryDee

Originally posted by: porkster

The THG clearly showed the AMD X2 isn't the winner claimed in other site's solo thread tests.

Anyone that monitored the THG stress and multitasking bench test, would have noticed the Intel was hard to match when running correctly.

The overall stress test was a botch up, with THG being such lamers and not using new thermal pad, the wrong fan and the wrong fan connector on the motherboard. Also once they got the Intel system setup properly they didn't reset the score of the multitasking challenge making the whole event a waste of time.

The overall picture that comes out the THG test is that the AMD X2 is a poor multitasker and not much a match for the top end Intel desktop CPU's, but probably a good upgrade path for those already with a mobo that are AMD based.

The price of the CPU's and their over ability to do current applications, doesn't make them very relative for customer's needs. Even the rich would be stupid to waste the cash on these CPU's with out a real need for the grunt.

Lets hope some other sites can test both CPU's in the future with a similar real test that puts the strain on the systems.

.

You've been repeating these things over and over, while disregarding other's views to the contrary.

When faced with logical opposition you fall silent for a while and then return repeating the same nonsense. Did you think we were going to believe it this time?

Do you think we'll believe it when you repeat it again in a couple hours?
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
Originally posted by: justly
Originally posted by: Duvie

...INtels performance is hard to get a read on cause it changes so much....first 3 hours it does 36fps in Farc cry and then the next day or so it only does 27 and now back to 35-36...I find that suspicious....CD encoding takes off then once it has a decent lead it slows down and speeds up the Far Cry....I would love to see if Toms is in their manipulating the data...Notice how the divx was running at 250:10 ratio in encoding time then it slowed all the way down to like 110:10...How convenient as it allowed cpu cycles to free up to help CD encoding and far Cry....

Duvie, you have more experiance with these types of apps than me, so I was wondering if you think this could have anything to do with how HT shares cache?
Could it be, as the treads get issued 2 heavily cache dependant apps could be assigned to different virtual cores that share the same cashe thus slowing them down (slightly) and allowing the other two less cache dependant apps speed up while shareing the other cores cache using its two virtual cores, then swithing around and causing the slower apps to speed up and the faster apps to slow down.

Other than your conspiracy theory about Toms this is the only explaination I can think of.


You are likely more right then my conspiracy theory...HT shares cache on it one core but we know that the INtel chips also share cache for both cores as a point of design..

While I can see your point...the run lengths of th eapps whether it be Farcry Winrar or even cd encoding the number of these things that are completed hourly makes it not quite jive with the trends I see that happen for longer period (sometimes 12 hours plus)......

Usually winrar held, cd encoding slowed down and Far Cry picked up...Then CD encdoing picked up and Far Cry slowed down to 27fps where it has sopent a longer period of time thus was its behind still 280+ runs (some of it due to downtime but not nearly all of it).....

Lately I save seen a major slowdown in Divx (which was not seen at all in the 1st round of testing) and it has translated in faster CD encoding, faster Far Cry and Winrar held true....I just find the randomness a bit odd...I think over a prolonged period of time a pattern still would have likely emerged....CD encoding, winrar arcives, and Far Cry runs are all pretty constant without too mucjh variable so a more consistent pattern should be visible.....

I always felt in my testing my HT was qute consistent...so it seems hard for me to understand this flopping...so one could see my conspiracy theory on this...
 

Pabster

Lifer
Apr 15, 2001
16,986
1
0
Originally posted by: Duvie
I always felt in my testing my HT was qute consistent...so it seems hard for me to understand this flopping...so one could see my conspiracy theory on this...

Keep in mind that the HT behavior might well be different on the Smithfield dual cores than it is on Prescott. It is entirely possible that Intel has made a few revisions to it.

That said, I'll hope your "conspiracy theory" is simply that Tom has made a fool of himself and that the results are invalid. Considering Tom has 99% AMD advertising now I find it hard to believe he has any real incentive to hand Intel anything.
 

MDme

Senior member
Aug 27, 2004
297
0
0
Duvie has a very valid point. Even if there were changes to the smithfield core to improve HT, it can never account for the near "telepathic" abilities that the 840EE is showing right now. I mean, it simply "knows" where to put the processing power to catch the X2 system. Even with architectural improvements to the smithfield core, all the 4 processes are constant, and the core is one and the same, a pattern has to emerge......right now the only pattern emerging is the 840EE spending more times in the tests where it is behind the X2.
 

justly

Banned
Jul 25, 2003
493
0
0
Duvie, thanks for your opinion about the cache. You where the only person I could think of (and trust) that did enough tests of Intels HT to have possibily seen this before.
I think your right that no pattern is evedent but then again (if there is a pattern) it might be so long that it is not easy to recognize.
 

Pabster

Lifer
Apr 15, 2001
16,986
1
0
On another note, I find it rather pathetic Tom still can't figure out how to cool off the P4. The average temperature is better now (around 68C) but that is still too high. It seems the fan is only averaging 3500rpm whilst the AMD rig's fan is >4500rpm. What gives with that? Also (perhaps coincidentally with the temperature drop) the Intel rig's performance curve is nearly even now instead of jagging up and down wildly. But I still assert that 68C as an average temperature is too high. He needs to get it down around 60C.
 

carlosd

Senior member
Aug 3, 2004
782
0
0
Originally posted by: cscpianoman
Porkster, you are quite correct that THG has set all the priority settings the same and started the programs in similar fashion. What you fail to realize, this is your downfall, is that priority setting are encompassed in the operating system. Hence the name "operating system." Let me walk this through with you. A processor is just what the name implies it processes or crunches numbers. The operating system sits on top of the processor and feeds the processor the information it needs to crunch. Any threads sent is the responsibility of the OS, not the processor. This can be amply verified through several methods. One, take a look at microsofts website and data sheets. The information is there, do your research. Two, there is a method of changing priorities in the task manager. Change the priority and the task will either be given preference or denied. However, this is a direct setting adjusting the operating system, not the processor.
Now I come to the current test. Intel is walking away with Divx, why? According to the OS there is 4 processing units running all the threads given them. There are four applications. Essentially, the OS is capable of giving the Divx thread to a "processor" because there is one availble. Let's now turn to the AMD processor. According to the Windows OS, there are two processors. Ahh, here we have an issue, because there are four applications. The operating system next takes a look at what has the highest priority and sends this task to a processor, same goes for the second thread. The third thread is stuck in limbo waiting for a process to finish, or if two threads have equal priority they will be switched back and forth allowing for an equal chance of processing. Now we come to the stickler, the fourth thread. We have already come to the conclusion through numerous posts that Divx is given a lower priority and that THG did nothing to change those settings. This lower priority setting is viewed, "by the OS," as being less important than the other three. Hence the OS doesn't send the thread unless there is room for it to fit. The processors, however, are running at capacity, ie 100%. There is no room to fit this extra thread, except at times when the loops break. This explains the reason why the AMD processor is "avoiding" the Divx thread because the OS isn't sending it. Now my speculation would be that if the priorities were equiliberated we would see a more equal result among all threads and processors.
I have layed all this out in very simple terms and from information gleaned from various sites including THG and Microsoft. I don't want your opinion, I already know what it is. This is for your information and to come to a more knowledgable understanding that when comparing AMD and Intel they both have their strengths and weaknesses. I am not trying to favor one or the other. This test demonstrates that Intel is very capable of doing four tasks at different priority levels, while the AMD processor focuses on three dissimliar priority tasks. I will address the other concern about wattage and energy efficiency later.

Excellent posts. Something that some stupid intel fanboys have to know before throwing crappy comments.
 

cbehnken

Golden Member
Aug 23, 2004
1,402
0
0
Originally posted by: Pabster
On another note, I find it rather pathetic Tom still can't figure out how to cool off the P4. The average temperature is better now (around 68C) but that is still too high. It seems the fan is only averaging 3500rpm whilst the AMD rig's fan is >4500rpm. What gives with that? Also (perhaps coincidentally with the temperature drop) the Intel rig's performance curve is nearly even now instead of jagging up and down wildly. But I still assert that 68C as an average temperature is too high. He needs to get it down around 60C.

Did you ever dream that maybe the Intel system is controlling the speed of it's own fan?

How again is this tom's fault? Do you believe he shouldn't use the fan supplied by intel and just put a 9500 RPM fan on so the 2 prescotts can stay cool?