Stress Test MK II

[Stimpy]

If they calculated the amount done per hour in the other processes with HT switched off, they would see that they actually get a performance improvment at the expense of Divx.
HT is not good if you have a low priority app hindering the performance of higher priority app.

I would like to see them running 4 threads of equal priority on both CPU's. Only then can we make a real judgement as to which processor is the best at multitasking

 

[Weescotty]

Update 8 - Talk about a gloss over of how the P840 got it's butt kicked.
Notice there was no screeny showing the results this time.

Now they are going to test single apps!!!!
WHY????
If we're going to run single apps only we wouldn't buy a P840 or X2.
What a bunch of jokers.

Bet if any of them bent over you'd see the shoes of an Intel tech hanging outta their ass, and vice versa.

<P ID="edit"><FONT SIZE=-1><EM>Edited by weescotty on 06/16/05 11:26 AM.</EM></FONT></P>

 

[Starfishy]

How did you calculate work done per hour? I don't think it is that easy, try to keep in mind that these tasks may not all be weighted the same in terms of CPU needs.

 

[Starfishy]

Actually, I am kind of pleased that they are going to check single app. If I was going to buy one of these CPU's for my normal multi tasking needs, I would still like to know that when I am running a game as a singe task what CPU will give me the edge. I may be doing alot of multi tasking, but I am certainly not multi tasking all the time. And if it would mean a great performance increase for my gaming, I would make sure to shut down my other apps. I'm not going to get to upset about this one, I'm glad they are doing it.

 

[Guest]

THis remains weird! Its like they d ont want to admit their own stupidity or they dont want to make the P4 look bad. Or both!

They dont want to up the priority while leaving the HT on. We either get low priority HT on or normal priority HT off.


Asus P4P800DX, P4C 2.6ghz@3.25ghz, 2X512 OCZ PC4000 3-4-4-8, Leadtek FX5900 w/ FX5950U bios@500/1000, 2X30gig Raid0

 

[Weescotty]

What else ya expect.
Already shown that with HT off the P840 gets its silicon butt owned by the X2.

Only way to make the P840 look good is to re-enable HT and leave the DivX priority set as low. At least they know that way the P840 will win at least the DivX test.

As for testing single apps - Sorry POINTLESS, if I was running single apps I wouldn't be after a dual core processor. Geez THG have gotta be the dumbest aholes in the review community.

Whats next? They'll test the original doom in DOS?

Gotta love this gem from update 8-
"Recall that our overall goal is to determine which platform is best-suited for a variety of uses"

Silly me I thought the original goal was a stresstest, but wait, the P840 with SLI bombed out, so lets change the goal now.

<P ID="edit"><FONT SIZE=-1><EM>Edited by weescotty on 06/16/05 12:17 PM.</EM></FONT></P>

 

[P4Man]

>I don't think Tom is coming back

I read somewhere he cashed out for a rather incredible ammount of money (was it $/€ 30M ?). I wouldn't come back either

Anyway, Tom was not exactly a guarantee for quality either; if THG was head and shoulders the best review site out there many years ago, it was simply because there was no serious competition. THG has simply maintained a status quo, whereas sites as Anandtech have become much more professional. Really, read some 1998 reviews by him, and have a laugh.

>THG will keep generating money despite getting worse day by
>day.

It certainly won't die overnight, but reputation is extremely important in the long run. Trust me, we aren't the only ones laughing at these clowns.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[HansGruber]

If you want to advertise a product that is for enthusiast, you are doing much better somewhere else than in THG.

But if you are selling magic magnets that reduce power consumption of cpu by 35%, then THG is good place to advertise. :lol:

<font color=red>"We can't solve problems by using the same kind of thinking we used when we created them."
- Albert Einstein</font color=red>

 

[Starfishy]

If I was a user who needed a rig for business multi tasking but was hoping to do some gaming on that rig as well (to save me buying two rigs) I would be very interested in how these CPUs handled single apps. I am not defending THG as we can all see that they have screwed more than a few things up here, I just don't think that testing single apps is that bad. I think that many people who were interested in buying these might at one point ask themselves how they perform at single apps.

 

[Weescotty]

Well -
Look at results for comparable single core CPUs. These dual core should benchtest somewhere in the same area given the app or game doesn't support multithreading.

Just think it's a waste of time testing single apps unless they support multiple threads. Plus single apps on both these CPU's have been tested and are reviewed on other sites.
This was the first stress test that had been tried.

 

[Guest]

Recall that our overall goal is to determine which platform is best-suited for a variety of uses

I gotta agre thats a freakin good one. Skipped over it.
That absolutely means nothing. Both platform are best-suited for a variety of use!Depends wich variety...

Asus P4P800DX, P4C 2.6ghz@3.25ghz, 2X512 OCZ PC4000 3-4-4-8, Leadtek FX5900 w/ FX5950U bios@500/1000, 2X30gig Raid0

 

[Huh]

Why are they using a bar chart to track load balance. The only thing chart tells you is loading at particular instance in time.

What makes more sense is a XY chart like the one they use to track the system temperature. This way you can see the trends in load balancing. If one application's load decreases, what other application(s) pickup the slack to bring the CPU usage back to 100%. Since Far Cry is the primary program is it always the highest load? Does that load change over time?

 

[Starfishy]

Has the X2 been running DivX better? Or has the EE just been slowed down without HT on? Or a combination of both? Can we even tell with the information that they provide us?

 

[HansGruber]

At last Intel is working properly, since idle apps won't steal cpu time from higher priority applications.

<font color=red>"We can't solve problems by using the same kind of thinking we used when we created them."
- Albert Einstein</font color=red>

 

[Weescotty]

Far Cry was up on the P840-
As of 8.24pm EST P840 is losing ALL 4 tests.

Unfortuneately without the CPU usage/Temps stats etc for the P840 it's impossible to know whats going on.
Of course this gives THG just what they wanted, they can now feed us any BS, and we have no way of knowing if it is true.

<P ID="edit"><FONT SIZE=-1><EM>Edited by Weescotty on 06/16/05 08:31 PM.</EM></FONT></P>

 

[theholylancer]

what is going on with the AMD load??? its gving 0 to winrar or farcry or lame once in every while........

 

[Weescotty]

Who cares lol.
Everything they have done to try and make the P840 look good has backfired, it looks positively OWNED by the X2 now.

 

[HansGruber]

Is that HT really so good ?

<A HREF="http://www.anandtech.com/IT/showdoc.aspx?i=2447" target="_new">http://www.anandtech.com/IT/showdoc.aspx?i=2447</A>

Looks like it sucks at Linux servers.


<font color=red>"We can't solve problems by using the same kind of thinking we used when we created them."
- Albert Einstein</font color=red>

 

[HansGruber]

Just on Linux servers thats all cuz Linux knows how to distribute priorities properly.

To some extent that is true..

But i believe that main reason is that HT is just a hack.
It makes OS believe that there is 2 CPU's, so it's making scheduler wait less = smoother running apps.

Server with HT gives unneccesary attention to background tasks, by stealing CPU cycles from main task, and that leads to lower performance.


<font color=red>"We can't solve problems by using the same kind of thinking we used when we created them."
- Albert Einstein</font color=red>

 

[HansGruber]

If HT is just a hack then why didn't AMD have it in their Athlon64 huh?

Shorter pipelines or something..

But it's AMD patent.

<A HREF="http://www.xtremesystems.org/forums/showthread.php?p=915197#post915197" target="_new">Mace a post in xtremesystems</A> worth a read (i hope).


<font color=red>"We can't solve problems by using the same kind of thinking we used when we created them."
- Albert Einstein</font color=red>

 

[HansGruber]

What you're doing is trying to do is attack the P-D without HT enabled

Not true. If im attacking anything it's HT enabled servers and multicpu systems with HT.
It's good for single core but not for servers.

Currently that P-D system without HT is executing code as OS and programs wants it to, which is the correct way.


<font color=red>"We can't solve problems by using the same kind of thinking we used when we created them."
- Albert Einstein</font color=red>

 

[HansGruber]

Get it straight, it ain't a hack and AMD ownz [-peep-] on SMT patents.

It offers NO performance advantage in real life situations that servers face.
It's a AMD invention but they don't use it.

There tons of other SMT patents by others that Intel have bought to use them.

What is the point ?
Are you trying to say that Intel provides all the technologies that AMD uses ?

If it wasn't for the heat problem I'd recommend 8xx (EE or non-EE) to anyone any day for the reason that it's cheap, suits real multitaking and not for gaming which is what the majority of dual-core/dual-chip users do like me.

By all means do what you like.
But don't try to say that Hyperthreading is useful in servers.

<font color=red>"We can't solve problems by using the same kind of thinking we used when we created them."
- Albert Einstein</font color=red>

 

[HansGruber]

As I've said before this is Pentium-D we're talking about here not servers.

I was talking about hyperthreading.
It is not any good in any system that has more than 1 CPU.
Surely you realise that HT improves some applications performance at the expence of other apps ?
What good is in that ?

It's only good when speed is more how you FEEL it instead of actual performance.
Situations like normal desktop user often has.
It really doesn't matter if HT steals 0.2 seconds from some encoding app when you open a browser, you wont miss that time.
But if you have to wait 0.2 sec longer you feel that computer is slow.

Multi CPU systems can handle that anyway, so HT only slows performance.

Does this help to understand ?


<font color=red>"We can't solve problems by using the same kind of thinking we used when we created them."
- Albert Einstein</font color=red>

 

[P4Man]

Oh please....

>It offers NO performance advantage in real life situations
>that servers face.

You can't conclude that from just ONE benchmark. It doesn't benefit MySQL read performance on small databases, that doesn't make it true in general. From a previous Anandtech database test:

Our tests have shown that Hyper Threading improved performance 3 - 5% on average

http://www.anandtech.com/IT/showdoc.aspx?i=1982&p=4

Its simple: sometimes HT improves performance, sometimes it hurts, and sometimes it matter nothing.

>It's a AMD invention but they don't use it.

*sigh*. Myths can be hard to kill sometimes. No its NOT an AMD invention, in fact the concept is at the very least 20 years old, and has been implemented in chips long before AMD filed that patent. Furthermore, the patent you are referring to doesn't even cover SMT like used in the P4, its about processing (hardware) interrupts in parallel with executing a software thread.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[HansGruber]

First came HT and applications did loose performance 10-20%, then apps got optimised for it..
Optimised software was compared to applications that lost performance because of hyperthreading, and those optimised apps increased performance 5-10%.. impressive..
:lol:

<font color=red>"We can't solve problems by using the same kind of thinking we used when we created them."
- Albert Einstein</font color=red>