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

[Duvie]

Originally posted by: Furen
Let me just say this: I love Hyperthreading, that is even though I prefer AMD CPUs just because... (I'm somewhat of a fanboy but not to the extreme, I'd say). Many times over the past few years I've been tempted to go Intel 'cause of HT because I've actually had use for multithreading and because I hate how XPs crap out when multitasking.

Now, I find the comparison between the ee840 without ht (a D 840, basically) and the $1k 4800 a bit unfair... they should have dropped the multi on the amd system to 11 to simulate the 4400, since this cpu would be closer to the price of an 840.

Now, I'm sure the 4400 would still destroy the 840 but it'd be nicer to make things a little bit more fair and this would settle the "performance" side of things on the more affordable cpus, which are probably what most people will buy, anyway.

What would be fair is if a bunch of imbeciles didn't run the test....what would be fair is if a bunch of people acting subjective but with underlying biases didn't run the test....I suggest the evidence is and was selection of 4 cpu test, use of Divx low priority setting in Gknot, and leaning to more encoding test (areas that even now are still the closest area Intel comes to X2)...why 2 encoding out of the 4...why not 1 encoding type of app and 1 rendering type of app??? Well that would have favored AMD which is and has even with single core cpus dominated in this area...

What they should do is run the test and run 2 of the test, maybe a couple different orientations of the 4 they use now....run it shorter length of time...2 hours is sufficient to get the idea or pattern....Then run 3 apps...then run 4 apps...then run 5 apps....then draw some conclusions...

The problem is these scam artist new exactly what they were doing...running a dual core versus a dual core with dual virtual cores....Now it made sense to properly load the PD EE fully to make sure all cpus (including logical ones) are maxed as a stress test component. BUt to use exactly 4 apps....Why not 5 or 6 or 7???

You can run 2 apps maxing all 4 cpus...If a person ran a gaming app and a rendering app like 3dsmax7 or Povray 3.7....You can also do it by running TMPGenc and those same rendering apps....just use 2 multithreaded apps....You can do it with 3 as well...The fact it was exactly four is the question....

The fact they used 3 apps of normal priority and then chose a program with low priority and then act fvcking dumb like they dont know "why the AMD seems to be "lagging""...Bunch of scam artist or ppl too stupid to have a hardware review site. what were they sleeping the last 2 years with HT and the northwoods???? They should run the test again and this time use a program like...adobe premiere, canopus procoder, TMPGenc, Mainconcept, Pinnacle, etc..countless apps that encode but do it at a normal priority so that all 4 apps are on equal footing when it comes to thread scheduling...

Then ofcourse you have Divx..think this was a random choice as well??? I think not...Divx has favored Intel cpus more so and in the X2 testing is th closest the intel dual cores come.....Divx to me seems to be fading..Most ppl seem to encode straight to Mpeg2 nowadays with ample DVD burners or use XVID...BUt you cant use XVID cause it favors AMD in all of the testing.....

He uses a substandard cooling on the Intel cpu....Guess what he did as well on the AMD X2...the HSF he used is not the retail bundled HSF but instead an older all aluminum cooler with a small 70mm low profile fan....newer HSFs use 80mm higher profile fans and a copper core HS with heatpipes..can you imagne how much cooler the X2 would have been running versus the 54c (which was about 13-14c below Intels temps) it was reporting during the test???


one cannot ignore all of these things.....There was ample bending over backwards for the troubled intel system to only later dismiss it like it was nothing.....


I do like the making it a 4400+ speed but like you said it would have not made any differnce in this last test....

 

[boran]

and THG sinks lower and lower into Intels pockets apparently, tbh I never believed before THG was on Intels payroll, I do very believe it now. that or they are really the biggest morans that are allowed to play with CPU's. they really do not understand a damnd thing about their own numbers.

I quote from their 8th update: (link might be referring to later updates in the future me thinks)

This also demonstrates that when numerous CPU-intense applications must run in parallel, Intel's hyperthreading can deliver immense performance boosts.

(Second paragraph below the ad)

which is complete and utter bullsh!t

and since I was bored I occupied myself with prooving that it is complete and utter bullsh!t

here you can find a lovely screenshot from an exell worksheet I made (it looks lovly I know) where you can see what everyone here already knew, intel is faster at DivX encoding with HT enabled at the cost of all other applications something you do not want to have when it is set to lower priority.

if you ever wanted to mass encode CD's, compress some files, encode a movie AND game, you will have the best gaming experience on the intel with HT disabled.

tom claims the DivX thread is not set to a lower priority but the numbers speak for themselves. as mentioned before most divX encoders will default encode at a lower priority.

and I think my numbers make very clear that intel now has become much faster at everything else, so disabling HT makes the intel proc into a thing that can just nip the heels of the AMD proc, wheras enabling HT makes it into a chip that owns AMD in one app and gets owned in the three others. personally i'd prefer the overall strong performance.

and once again before the intel fans go ballistic on me, the faulty behaviour with HT does not lie with the intel proc, it does not schedule its ownthreads, the OS does that and clearly winXP is not 2xHT aware.

 

[imported_Frallan]

Duvie sums the Bias issues up pretty well but he(/she, sorry dunno) forgets another beauty a@ thg stress test.

They measure the amount of power the systems draw with a nice little gizmo, but since they + world (incl. dogs) know about the difference in power dissipation (sic) between the Intel and the AMD solution they measure them together. Which to me as a consumer is totally irrelevant. Worst case scenario (if they cant find another gizmo) switch the Gizmo between the system when reseting the scores.

 

[mircea]

Originally posted by: Furen
Now, I find the comparison between the ee840 without ht (a D 840, basically) and the $1k 4800 a bit unfair... they should have dropped the multi on the amd system to 11 to simulate the 4400, since this cpu would be closer to the price of an 840.

I agree that a CPU vs. CPU would be more fair, but if you start pointing out prices, think about the fact that if I would buy the X2 I'd pay the price for the CPU only, as I and many already have a 939 mob. And I have a very capable cooler so I could go OEM.
If I want to go to Intel dual solution, then I'd be forced to buy a new MOB, CPU and RAM, and can't go OEM (not a multi-socket cooler 😉 )

Anyways, at least now we see we were right about the scheduler, and the HT is really good for what it was built. If I'd have the money I would go for the HT enabled Intel since when I'd be in a position where HT hindered my work, I could disable it and enable it again when it would be beneficiarry.
But I can't afford to get a lower performing system (PD non HT) for the same money I could get one X2, nor can I get a super HT enabled system when that would cost me $1500 just for the Mob, CPU and RAM. Again I and many are faced with changing not only CPU's but other components as well so price/performance is not so equally as other times.

 

[Duvie]

Originally posted by: Duvie
Some monitoring of the data over 2 days (latest round)


Time-------Lead (Lame)/Co.--------Lead (WinRar)/Co.--------Lead (FarCry)/Co.-------Lead (Divx)/Co.

Wednesday

7:05am---------18 (AMD)---------------175 (AMD)----------------47 (INTEL)---------------40min (AMD)
7:35am---------18 (AMD)---------------180 (AMD)----------------47 (INTEL)---------------40min (AMD)
8:38am---------19 (AMD)---------------189 (AMD)----------------45 (INTEL)---------------40min (AMD)
9:34am---------20 (AMD)---------------198 (AMD)----------------46 (INTEL)---------------40min (AMD)
10:20am--------20 (AMD)---------------205 (AMD)----------------45 (INTEL)---------------40min (AMD)
11:39am--------21 (AMD)---------------217 (AMD)----------------43 (INTEL)---------------40min (AMD)
12:41pm--------21 (AMD)---------------227 (AMD)----------------44 (INTEL)---------------40min (AMD)
2:55pm---------26 (AMD)---------------249 (AMD)----------------48 (INTEL)---------------50min (AMD)
6:14pm---------26 (AMD)---------------280 (AMD)----------------43 (INTEL)---------------50min (AMD)
7:03pm---------27 (AMD)---------------288 (AMD)----------------43 (INTEL)---------------40min (AMD)
7:56pm---------28 (AMD)---------------297 (AMD)----------------43 (INTEL)---------------40min (AMD)
9:12pm---------28 (AMD)---------------310 (AMD)----------------43 (INTEL)---------------40min (AMD)
10:01pm--------29 (AMD)---------------316 (AMD)----------------42 (INTEL)---------------40min (AMD)

Thursday

7:01am---------31 (AMD)---------------398 (AMD)----------------41 (INTEL)---------------50min (AMD)
5:01pm---------34 (AMD)---------------477 (AMD)----------------16 (AMD)----------------60min (AMD)
6:09pm---------34 (AMD)---------------487 (AMD)----------------23 (AMD)----------------60min (AMD)
6:31pm---------35 (AMD)---------------489 (AMD)----------------25 (AMD)----------------60min (AMD)
8:28pm---------35 (AMD)---------------505 (AMD)----------------37 (AMD)----------------60min (AMD)
9:19pm---------35 (AMD)---------------511 (AMD)----------------42 (AMD)----------------60min (AMD)
10:31pm--------35 (AMD)---------------523 (AMD)----------------49 (AMD)----------------60min (AMD)
11:35pm--------35 (AMD)---------------532 (AMD)----------------56 (AMD)----------------60min (AMD)

Friday

12:37am--------35 (AMD)---------------542 (AMD)----------------62 (AMD)----------------60min (AMD)
1:34am----------36 (AMD)---------------549 (AMD)----------------68 (AMD)----------------60min (AMD)
2:21am----------36 (AMD)---------------555 (AMD)----------------72 (AMD)----------------60min (AMD)



Too bad I dont have the 10hours earlier today when the tide changed in FarCry...

Total time of data......43.5 hours (***All numbers are for AMD which currently is increasing its leads in all of them***)

--------------------Total------------First 5 hrs-------First 15 hrs-------Middle 9 hours----Last 19.5 hours------Last 5 hour

Lame ------------ 0.41/hr-----------0.60/hr---------------0.73/hr---------------0.22/hr-------------0.26/hr-------------0.2/hr

WinRaR---------- 8.74/hr-----------9.00/hr---------------9.40/hr---------------9.11/hr-------------8.05/hr-------------2.97/hr

FarCry(runs)---- 2.74/hr-----------0.60/hr---------------0.33/hr---------------0.11/hr-------------5.79/hr-------------6.00/hr

Divx(min)------- 0.46/hr-----------0.00/hr---------------0.00/hr---------------1.11/hr-------------0.65/hr-------------0.00/hr

 

[Sixtyfour]

Originally posted by: mircea
HT is really good for what it was built.

To me it looks like HT isn't any good, it does thing that you don't want it to do.

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.

Originally posted by: boran
intel is faster at DivX encoding with HT enabled at the cost of all other applications something you do not want to have when it is set to lower priority.

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%..

😀

 

[boran]

Originally posted by: Sixtyfour
Optimised software was compared to applications that lost performance because of hyperthreading, and those optimised apps increased performance 5-10%..

😀

We're talking virtual quad core here eh. and the optimizations here wouldnt do anything, it's the OS scheduler that's unable to handle 4 cores. and imho MS has quite a task set out for them, because CPU load on consumer multiproc PC's will be much more variable than CPU load on server/workstation PC's. so the work they would have to put into the scheduler would actually be more for the home OS than for the server OS

hell, my head already starts spinning when I have to program an app with over 2 threads, let alone 4.

 

[Sixtyfour]

I mean HT advantage in general.

First loose 10-20% and then get half of it back after millions spent in coding. 😀

 

[Pabster]

Originally posted by: Gamingphreek
The AMD chips simply slaughter the Intel chips. And that is under the greatest stress you can give a CPU. This shows how Toms test should have been (of course Opterons and Xeons instead of P-D's) and a server environment.

Funny how your fanboy bias taints any credibility you might have had.

Slaughters? I'd hardly call it a slaughter. The Nocoma core Xeon (with HT) are very competitive with the Opterons. And the title of this thread (in case you haven't noticed) was X2 versus P4EE, not Opteron vs Xeon. Why do the AMD fanboys always start bringing up Opteron? We all know Opteron is great performance-wise, but the price tag kind of puts it out of reach for the desktop. We're not discussing 64-bit linux database servers in THIS thread, so move on.

We have been telling you this entire thread that while intel chips are still good they are simply trumped by AMD's chips. This is an undeniable series of results.

Ok, so because the Opteron-powered servers can cough up a few more page requests per second, that means AMD is the be-all to end-all? Come on, give it up. If you want to be a blatant fanboy do it somewhere else.

 

[imported_Frallan]

Originally posted by: Gamingphreek
Link

You can start apologizing any time now (Porkster, Pabster, Dothan). That is how a real test suite is performed. All stability issues were noted as well.

-Kevin

Sorry to say gamingphreak but that ^^ was just Trolling. Try to stay above the Porkster level please.

And right now:

36 582 90 70

Divies last stats was:

36 555 72 60

 

[mircea]

Originally posted by: Sixtyfour

Originally posted by: mircea
HT is really good for what it was built.

To me it looks like HT isn't any good, it does thing that you don't want it to do.

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.

I feel that you don't realize what HT was made for. It was made not soo much to improve multitasking as it was a way for Intel to coverup the holes they made in the netburst architecture, by giving the processor work to do to fill up those long pipelines. As a proof of that just look at the Pentium performance now. It's the same exact 2 cores but the OS is giving it work as it does to any CPU. Because of the long pipelined netburst architecture the Pentium is now starved, it does less work per cicle. So it's not really taking off on the other 3app, that DivX was "stealing cicles" since those stolen cicles are now used fully.

HT does it's job perfectly.
The Windows operating system is the one that makes HT hinder aplication when it doesn't know that the lower level priority should not get equall share of work on an HT enabled system.

 

[Gamingphreek]

Originally posted by: Frallan

Originally posted by: Gamingphreek
Link

You can start apologizing any time now (Porkster, Pabster, Dothan). That is how a real test suite is performed. All stability issues were noted as well.

-Kevin

Sorry to say gamingphreak but that ^^ was just Trolling. Try to stay above the Porkster level please.

And right now:

36 582 90 70

Divies last stats was:

36 555 72 60

That was in no way a troll post. I quoted information. However Pabster cannot get his head out of his a$$ long enough to think about what he is reading. Try not to be a noob, youll get farther. (Junior member calling diamond member a troll).

The Nocoma core Xeon (with HT) are very competitive with the Opterons

Oh well screw the irwindales then which are the noconas successor. I can also tell you didn't read the article at all because in a lot of cases HT decreases performance.

X2 vs PEE is essentially the same as Opteron Vs Xeon. What do you think the Irwindale Xeon is, pretty damn similiar to the PEE. THe Opteron is merely an X2 which requires registered DDR and has a few other server grade enhancements to the core.

but the price tag kind of puts it out of reach for the desktop.

Because the xeon isn't :roll:

We're not discussing 64-bit linux database servers in THIS thread, so move on.

Why not. It seems completely relevant. You say, Intel is the best out there unjustified. If you say you prefer intel and backup your reasons no one would have any problem.

If you want to be a blatant fanboy do it somewhere else.

I am not a blatant fanboy. I was running Intel chips for the longest time. My dad and i even have a little contest going on for fun as he has Intel. His 2.8C slaughters my 2500+ in most everything (if not everything). I didn't buy the northwoods at the time because the AMD chips were so much cheaper. I built my first computer at age 10 (17 now). At which time while Intel was the undisputed performance leader, AMD was much cheaper, i could not afford intel. I would have loved to buy a northwood 2.4C and OC clock it, but i simply couldn't afford it at the time. I have even used Cyrix many times. I have no problem with any company, and will buy the processor that suits my needs the best.

-Kevin

 

[Pabster]

Originally posted by: Gamingphreek
That was in no way a troll post. I quoted information. However Pabster cannot get his head out of his a$$ long enough to think about what he is reading. Try not to be a noob, youll get farther. (Junior member calling diamond member a troll).

You are trolling. And you still haven't told me what I should "apologize" for. Now get your head out of your rear and either stick to the topic of this thread, or get out. We can't have any discussions on this board anymore because fanboys like yourself always turn the threads in to a flame-shoot. And don't flatter yourself for being a Diamond Member (as I am equally) because that doesn't hold any muster.

The Nocoma core Xeon (with HT) are very competitive with the Opterons

Oh well screw the irwindales then which are the noconas successor. I can also tell you didn't read the article at all because in a lot of cases HT decreases performance. [/quote]

Actually, I did. Thanks for playing! I already knew that HT decreased performance in some cases.

X2 vs PEE is essentially the same as Opteron Vs Xeon. What do you think the Irwindale Xeon is, pretty damn similiar to the PEE. THe Opteron is merely an X2 which requires registered DDR and has a few other server grade enhancements to the core.

We're not discussing 64-bit processing here; We're not discussing Linux; We're not looking at the performance of database servers; We aren't comparing Opteron and Xeon. Those are SERVER CLASS processors with SERVER CLASS price tags. While architecturally they may indeed be (very) similiar to their desktop counterparts that is irrelevant to this discussion.

Why not. It seems completely relevant. You say, Intel is the best out there unjustified. If you say you prefer intel and backup your reasons no one would have any problem.

LOL! Where did I ever say Intel is the best out there, period? Show me. I have stated several times that *I* prefer Intel. Obviously a few people here have a problem with that.

 

[Gamingphreek]

You are trolling. And you still haven't told me what I should "apologize" for. Now get your head out of your rear and either stick to the topic of this thread, or get out. We can't have any discussions on this board anymore because fanboys like yourself always turn the threads in to a flame-shoot. And don't flatter yourself for being a Diamond Member (as I am equally) because that doesn't hold any muster.

Jeez what do i have to say to get this through your thick head. That is not a troll post. I quoted an article that is highly relevant to this discussion and analyzed it. What is so hard for you to understand.

Actually, I did. Thanks for playing! I already knew that HT decreased performance in some cases.

Then why did you say that the Xeon (With HT) was competitive. On these test HT hurts performance.

We're not discussing 64-bit processing here; We're not discussing Linux; We're not looking at the performance of database servers; We aren't comparing Opteron and Xeon. Those are SERVER CLASS processors with SERVER CLASS price tags. While architecturally they may indeed be (very) similiar to their desktop counterparts that is irrelevant to this discussion.

What is the difference!?!? They are archiecturally the same so that should reflect their desktop counterparts. What does price have to do with it. If we are using it to to provide a basis for their desktop counter parts then there is no difference, as price is not an issue there. You keep defending Intel and say they are "superior" (Yes, i know what you said, so dont try to play it off) however you cannnot provide any backup whatsoever to that statement. You call me a troll, yet once again you have no basis for that argument.

I dont know why I bother. You were hopeless from day 1.

-Kevin

 

[Lithan]

Opterons are cheaper then X2's or P4D's. And Intel loses in both 64 bit performance and in dual core performance. Even Tom's with all their bull**** can't put intel in the lead.


Frankly it looks like the ONLY situation Intel can win anything is with 4 apps running and one's priority below normal. Be proud intel fanboys. Intel can beat AMD *cough In 1/4th of the applications running in a situation that will never occur and if you specifically set up the amd to lose cough* Excuse me, must have a cold.

 

[Duvie]

WTH is he doing now???

He had stopped whole test around 6:00PST and now has recently shown just winrar working in live charts...Neither cpu is showing activity in cpu usage and I think those charts are not being updated currently....

Is he goimng to do 1 app per time then 2 apps, then 3 apps??? He must be reading this thread....

 

[Sentential]

Originally posted by: Duvie
WTH is he doing now???

LOL.... c'mon this is THG. Why are we suprised?

 

[biostud]

When using the top end solution all features should be turned on IMHO, but as we know all the normal P4 D, does not have hyperthreading enabled, which would make these chips very slow in this setup.

 

[Pabster]

Originally posted by: Lithan
Opterons are cheaper then X2's or P4D's. And Intel loses in both 64 bit performance and in dual core performance. Even Tom's with all their bull**** can't put intel in the lead.

What exactly have you been smoking? I don't recall Opteron being cheaper than X2 or P4 EE, and certainly not even close in price to the "D" series chips.

 

[Duvie]

Originally posted by: Pabster

Originally posted by: Lithan
Opterons are cheaper then X2's or P4D's. And Intel loses in both 64 bit performance and in dual core performance. Even Tom's with all their bull**** can't put intel in the lead.

What exactly have you been smoking? I don't recall Opteron being cheaper than X2 or P4 EE, and certainly not even close in price to the "D" series chips.

I agree they have huge premiums....200 series are meant to be 4 core systems and so they really sock the premium to you...The closest is the 100 series which is just a dual core opteron that will not work with another cpu....Even those prices as they have been announced are slightly higher then comparable X2 chip...

 

[boran]

Originally posted by: biostud
When using the top end solution all features should be turned on IMHO, but as we know all the normal P4 D, does not have hyperthreading enabled, which would make these chips very slow in this setup.

please see my post, HT doesnt create more CPU power it just redistributes it differently.

 

[Sixtyfour]

We all are right = all your base belong to AMD.

 

[Markfw]

Did somebody notice they manually set the AMD down to 1834 mhz, instead of 2400 !!!!!
And the Intel is still running at 3200 !!!

aholes at Toms ! what morons.

And the AMD went down to 48c (wonder why) and the Intel temp is disabled or broken. How bad can this get ?

 

[Duvie]

OK Toms...you are fvcking boring me...move on....NO SURPRISES HERE

Xbit labs ran Winrar and AMD scored 616kb/sec vs 478 kb/sec of INtel for a 28.9% lead

THG as of 5:05 had it scored 615 packed archives to 475 of INtel for a 29.5% lead..

Since THG doesn't normally agree with more reputable site, there should be some dancing in the streets on this one...

MOve to 2 app test you stupid germans!!!!! RUN them for 4 hours or less and try it with HT back on....

EDIT: not to be derogative to Germans in any way....Toms boys just happen to be stupid and germans...likely a coincidence...

 

[rgreen83]

Why the title change?