Stress Test MK II

Page 15 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Wow.. I never thought to look at the German site. They said the heatsink failed... probably meaning the fan and it is what Intel sent that was specifically designed for dual core (a much larger coper core and slightly faster fan). They also did replace the power supply with a new Enermax EG485AX-VHB 480w power supply. The system has been up now for a over 9 hours and holding strong.

So that is 6 boards, a heatsink and power supply, aand pull one graphics card... and the AMD system keeps running.
 
What the german site says is that the Intel mobo came with 2 hsf units. They originally tried running with the hs that had a smaller base, fewer fins and a slower fan. Now the are using the other hsf. It's just that the speed doesn't seem right to me. I dont recall seeing a chip with a 4200rpm fan for an Intel chip. On the other hand, if they upped the voltage on the other fan to 17v, they would get about that speed.
 
Who needs throttlewatch? Just look at the graphs. A couple times each hour, the P4 cools down, at the same time cpu usage drops right off. The Intel chip is throttled 25% of the time.
 
I doubt that is a sign of throtteling. Throttelig works a whole lot more subtle than that, furthermore, it wouldn't show up in the cpu usage graphs since the wait cycles throtteling introduces also keep the cpu busy as far as the os is concerned. A fully throttled cpu also shows 100% cpu usage. Those spikes are probably just some artifact from a benchmark finishing and restarting

= The views stated herein are my personal views, and not necessarily the views of my wife. =
 
>After all, Porky got tired of being raped by everyone here.
>LOL :)

Trolls don't get tired of being raped, instead, they love it. The only thing that tires them, is being ignored.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
 
Must be the divx benchmark then 😛 Since the same behaviour isnt showing up on the amd machine...

18 hours for the intel im impressed.

What happened to the temps though?

It ran at 90 c, they then changed cooler and it ran at about 75. Then i saw it run at about 70 and now its running sub 70? And you have to consider that hyperthreading does add a few degrees.

N e ways... Looks like its catching up with the cd encoding.

When it does run it's a mean one in this test.

PS: I'm a newbie... But does any of these chips switch off a core (idle the second core / cool&quiet) when you eg. just use word??? Would be a nice feature if it would switch the second core on and off if needed...

PPS: For me to find 2 cores usefull I'd like it to act like this:

Core 1: running windows and whatever programs I specify (eg: background AV, firewall, defrag program and or dvd / mp3 encoding, word and other low cpu usage programs).

Core 2: Off (due to power usage) but used on eg: games or rendering or whatever cpu intensive program that I myself would be working with...

Is something like this even possible with windows and the processors themselves?<P ID="edit"><FONT SIZE=-1><EM>Edited by NT78stonewobble on 06/09/05 07:26 AM.</EM></FONT></P>
 
wusy: Well the test does seem mightily intel suited and thg havent stated anywhere that running 4 (exactly 4 processes) is a typical userprofile. I cant imagine it is.

Multitasking is kewl and all but very few people can actually work on 2 things at exactly the same time. Even the brain idles tasks and thoughts in the background.
 
Its not that mean, its actually slower than the X2.
Consider this, the intel rig so far has completed:

188 encoded CDs
1013 winrar archives
660 far cry runs (@27FPS)
880 minutes of DVD encoding

Now, the P4 can run 4 hardware threads, (incorrectly) assuming they all run at equal priority, they should all get equal ammount of cpu cycles, which means, all these finished workloads represent an almost equal amount of cpu time. lets call that "single threaded P4 performance per 20 hours" or in short STPPs 😀 So the 840EE has achieved so far:

188 encoded CDs = 1 STPP
1013 winrar archives = 1 STPP
660 far cry runs (@27FPS) = 1 STPP
880 minutes of DVD encoding = 1 STPP
In total: 4 STPPs of course.


Now lets see what the AMD rig does:
1.04 STPPs on the encoded CDs
1.52 STPPs winrar archives
1.46 STPPs Farcry runs (at higher FPS too, ignored for now)
0.05 STPPs DVD encoding

In total: 4.07 STTPs versus 4 STPPs for the 840EE. They are neck and neck in troughput.


****BUT***

These apps arent running at equal priority, the DivX encoding is running at a lower priority (default of the app) AND farcry is running at higher priority (windows priorities foreground apps). that means, these weighings are NOT equal. The 840 spends more cpu cycles on farcry and less on the DivX encoding. How much more or less is anyones guess, but lets assume 50% for arguments sake. That means so far the 840EE has achieved:

188 encoded CDs = 1 STPP
1013 winrar archives = 1 STPP
660 far cry runs (@27FPS) = 1.5 STPP
880 minutes of DVD encoding = 0.5 STPP

In total, still 4 STPPs of course. Now lets see what the AMD rig does:
1.04 STPPs on the encoded CDs
1.52 STPPs winrar archives
2.19 STPPs Farcry runs (at higher FPS too, ignored for now)
0.02 STPPs DVD encoding

In total: 4.8 STPP versus 4 for the 840EE. The X2 is whipping the 840EE in throughput when you normalize the benchmarks.

And then there is the Farcry time demo issue. I'm not sure how I must read that, the X2 does ~50% more runs, and at a much higher framerate, does that mean it actually does twice as many frames, or is the higher framerate the reason in completes more runs ? I'll ignore this until someone can shed some light on this.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
 
A demo consists of eg: 1000 frames.

If you run at 100 fps a sec it will take 10 secs to complete that demo.

The amd being ahead is more pronounced due to the intel downtime i think. But still

Even if you run through the demo at 1 fps more then it adds up to a lot in the end...

PS: Why the rather complicated calculations with unknowns?
All you need is total run time for each machine then calculate how many cd-encodes, compressions, farcry-runs and divx encodes each one does per hour and then compaire the 2 machines.

That result would be true to the "test" theyre making...

I'd rather agree that the way the "test" was designed was flawed...
<P ID="edit"><FONT SIZE=-1><EM>Edited by NT78stonewobble on 06/09/05 08:43 AM.</EM></FONT></P>
 
>A demo consists of eg: 1000 frames.
>If you run at 100 fps a sec it will take 10 secs to complete
>that demo.

Thats what I thought, but the numbers don't add up then. From live charts:

Current Frame Rate on Intel System: 24 FPS at 680 runs
Current Frame Rate on AMD System: 36 FPS at 965 runs

That is a 66.66% lead in FPS and a 70.4% lead in #runs. If FPS was directly correlated to the #runs per day/hour, those ratio's should be indentical or at least within rounding error of the FPS, and they don't seem to be. We'll see what happens when the numbers get larger...


= The views stated herein are my personal views, and not necessarily the views of my wife. =
 
According to Toms last entry into the stress test it was not the EE at all but absolutley everything around it. they even replaced the PSU because they were sure it was going to be the next problematic part.
 
Judging by the impressive performance the P4 is currently putting in can we assume that it was heavily throttled before, when the temps were running over 85 degrees?


---
<font color=green>AMD</font color=green> Athlon64, Abit AV8.
<font color=blue>Intel</font color=blue> Dual PIII, Asus P2B-DS.
<font color=blue>VIA</font color=blue> Epia M9000.
<font color=red>I</font color=red><font color=green>B</font color=green><font color=blue>M</font color=blue> Thinkpad 570E.
 
>PS: Why the rather complicated calculations with unknowns?
>All you need is total run time for each machine then
>calculate how many cd-encodes, compressions, farcry-runs and
>divx encodes each one does per hour and then compaire the 2
>machines.

How else can you compare 300 more far cry runs with 600 more minutes encoded DVDs ? Sure, you could say 'chip X wins 3 tests and chip Y 1 test', but that isn't meaningfull when comparing throughput. My argument shows is the X2 is actually getting more work done overall without any unknowns other than the thread priority effect, but even if you leave that out, the X2 leads (by a hair).

= The views stated herein are my personal views, and not necessarily the views of my wife. =<P ID="edit"><FONT SIZE=-1><EM>Edited by P4Man on 06/09/05 08:50 AM.</EM></FONT></P>
 
P4 thats because the intel is behaving oddly. The cpu usage varies and so does temps.

The farcry test varies between 37 fps and 26.

Dunno wots it doing.

One could conclude all sorts of weird stuff from this test (great even more confusion for potential buyers :/):

Eg:

Intel -> you cant game and run other demanding processes due to fluctuating fps...

Intel -> Isn't stable with a vide range of graphx cards, mobo's, psu's and coolers...

AMD -> cannot perform when using 4 demanding tasks at once...

AMD -> is a gazillion times slower than intel at divx encoding...

And all of these conclutions would be wrong in a way...

PS: Again we need the total runtime for each of the systems. And you need to calculate in that the amd has had the whole test to run the programs and the intel only its uptime...

PS: I'm thinking you can't compaire farcry performance to divx performance at all...

Unless you know exactly how many calculations the cpu has to make to do 1 farcry run and one divx run. And then it would be again meaningless due to the fact that every single program is different from one another so you can't conclude anything either from just 4 programs...<P ID="edit"><FONT SIZE=-1><EM>Edited by NT78stonewobble on 06/09/05 08:57 AM.</EM></FONT></P>
 
> And you need to calculate in that the amd has had the whole
>test to run the programs and the intel only its uptime...

AFAIK, the AMD machine was also restarted, and the tests where restarted when they finally got the intel machine to work. I don't think the "uptime" indication == the time spent achievin the benchmarks. You can reset the benches without rebooting the machine.

now if indeed the AMD would have been running those benchmarks for a (considerable) longer time, my conclusions are as useless as the test itself, but I thought they both started again at the same time.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
 
o wait, it seems indeed runtime is NOT equal.. oh my, what on earth are we comparing then 🙁

= The views stated herein are my personal views, and not necessarily the views of my wife. =
 
If it doesn't throttle at 90°C, I guess it never will.
In fact, it seems clock throtteling indeed doesn't save a dual core P4 from falling over.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
 
Well the amd has been running 42 hours atm since the beginning of the test (or they resetted the test).

The intel has been running 20 hours since it's last reboot and its had 3 of them. So it's kinda hard to know how longs its actually been running the tests or down...

But then again shouldn't downtime actually be calculated in?

Then you could just use the amd number of 42 hours and 41 mins, if you do that then in this test each machine does this per hour:
Intel Amd
Lame encodes 4,62 | 4,64
Rar compressions 24,93 | 38,08
Farcry (demo runs) 16,10 | 22,83
DivX encodes* 21,54 | 0,94

* Minutes encoded per hour...

PS: P4man should anything like a cpu actually run at near 70 degrees. Just wondering my oc'ed xp-m gets unstable at 60+ degrees. Now I know the architechtures are very different but still????<P ID="edit"><FONT SIZE=-1><EM>Edited by NT78stonewobble on 06/09/05 09:19 AM.</EM></FONT></P>
 
>But then again shouldn't downtime actually be calculated in?

I don't think so. If you would, you'd have to excuse the machine for rebooting. Nah, stability and performance are two distinctly different issues, and you can't mix them and come up with a single figure/winner for both criteria combined. Thats nonsense.

> Intel Amd
>Lame encodes 4,62 | 4,64
>Rar compressions 24,93 | 38,08
>Farcry (demo runs) 16,10 | 22,83
>DivX encodes* 21,54 | 0,94

I'm not checking your math or numbers, but assuming these numbers are correct.. what do they mean ? Not a thing ! You can't just add those numbers to see who gets the most work done, thats why I cooked up my post. Now I do agree you should correct my numbers with a quotient that reflects respective run times, if we have that number..

= The views stated herein are my personal views, and not necessarily the views of my wife. =
 
As far as i know the 840EE DOESN'T have thermal throttling?@(Thermal Monitor2)enabled, since the EE version is for "Extreme users" who have good cooling.
BTW, it also lacks EIST which is useless anyway and the EE wont be compatable with the i945 chipsets.

The Pentium D CPU's do have EIST & TM2 enabled, so you should see those ones throtlling at extreme temps.
 
Yeah stability and performance are to different things "unless" you could conclude that in the timeframe of the 42 hours and *** mins a typical amd dual processor would run continiously and an intel would reboot 3 times and not due anything during its downtime.

Offcourse we cant do that so I just used the amd runtime number and used it for the intel.

The numbers show how much every machine "produces" every hour in each of the 4 tests. But offcourse calculating in the intel downtime would increase its numbers...

But youre right I'd doesnt show how much total work either intel or amd is doing.

PS: why am I even discussing some product I can't afford??? :S

@ blueswords ok didnt know that... But I still find it strange that cpu load and temps fluctuate on the intel processor and not the amd.

Perhaps the cpu running out of work at some points or waiting for another component? Eg: the write of divx to harddrive?<P ID="edit"><FONT SIZE=-1><EM>Edited by NT78stonewobble on 06/09/05 09:35 AM.</EM></FONT></P>
 
>PS: P4man should anything like a cpu actually run at near 70
>degrees. Just wondering my oc'ed xp-m gets unstable at 60+
>degrees. Now I know the architechtures are very different but
> still????

70°C is not *necessarely* a problem. Some chip are designed to work well over 100°C, it just depends if its designed for it or not.

Now I don't see the datasheet for the 840EE on intels site, but I so see intel defines a Maximum Tc (c="case", but of the cpu, not the ATX box) of 70°C for the Pentium D. If these temperature readings by THG are indeed accurate, and if the 840EE has a similar Tc (which I have to assume), it is indeed a problem, and it is running outside its specifications, which will almost certainly impact its lifespan, and possibly might affect stability.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom