AMD lying or not?

Balinju · 05-09-2003, 07:07 AM

I found this on amd's web page, is it to be believed???????????
According to them, the P4 3.06 hyper-threading enabled is slower than the P4 3.06 with the hyper-threading disabled, but further more, the amd 3000+ is better than the 2 by 17% and 10% respectivly.

Could this be actually posted on this thread ??

JSimmons · 05-09-2003, 07:20 AM

They were running the test with a 16-bit app?

pelle76 · 05-09-2003, 07:49 AM

I wonder if they really have gotten those HyperThreading scheduling algorithms right yet. If they haven't then the lesser result with hyperthreading on doesn't suprise me. I guess HT requires a little more overhead, and therefor without a proper scheduling algorithm the performanceloss is greater than the gain... ???

Balinju · 05-09-2003, 08:12 AM

Here you will find the benchmarks that they carried out.

Blackeagle · 05-09-2003, 12:25 PM

Balinju,

Didn't find any benches there. Mistake? Taken down?

Balinju · 05-09-2003, 12:45 PM

oops sorry i did a mistake Here you go . and actually it is not the benchmark, it is the software and hardware used for the tests, they do not show the results of each test. they just show the final result that is the graph i posted above

hara · 05-09-2003, 12:47 PM

That is just a benchmark in favour of AMD or a compilation of a set of (unpopular) benchmarks AMD chose in their favour. Intel also used to do it. Look here:
These are things that run in the business world.

Brad · 05-10-2003, 05:56 AM

they are just using select benchies, just like intel did when the P4 was first released

Rayman2k2 · 05-12-2003, 12:13 AM

well, by personal experience, the 3.06Ghz w/ HT ran somewhat faster than w/o HT, especially when the computer froze, and I was able to kill the frozen task with the "virtual" processor.

Balinju · 05-12-2003, 06:17 AM

i saw some benchmarkings of the 3.06ht and 3.06 w/o ht.
in rendering benchmarks the 3.06ht smoked the 3.06 w/o ht.
and also it smoked the 3000+ barton

but i still prefer AMD

Blackeagle · 05-12-2003, 12:07 PM

After reading this tread I did some looking around. I was concerned as I'm thinking on buying a XP3200+ once they are out and the price drops a bit.

At the Inquirer I read a article on this issue that make referance to a PCWorld set up benchmark made up of differant apps. that are in real world use, not synthetics (sp). Seemed a good idea to me, after all I'm not into benching.

According to PC World the XP3000+ does beat out the P4 3.06 CPU, just not by 17%. The 3000+ scored 137 in their benches, the highest ever scored on the best system tested. They ran the same tests on two other high end set-ups from other suppliers and they scored 134 & 136, close.

The p4 3.06 best score on these tests was a 132 (three systems also), while the worest was a 117 with the last system ran them at 127.

Comparing scores the 3000+ came out on top, in these tests. And for me the scores on a test that isn't a real app is less impressive than tests run on real apps.

ALL the differant top benchies these days seem to show bias. And it seems Intell is by far the big winner in that bias. So for me tests makeing use of real apps are the ones I'll go on. I think a comparison of past benches and present ones show this to be a real problem. But that has been covered by many others better at judging than I already.

I know this won't matter to those who love to try to set new high marks in 3DMarks or other benchies. They have to go with what the benchies are, just the way it is.

05-09-2003, 08:12 AM	#4
Balinju Cooling Savant Join Date: Dec 2002 Location: Malta Posts: 495	Here you will find the benchmarks that they carried out. __________________ So the bullet proof vest aint a $hit when d laser is pointed to your head Kid

05-09-2003, 12:45 PM	#6
Balinju Cooling Savant Join Date: Dec 2002 Location: Malta Posts: 495	oops sorry i did a mistake Here you go . and actually it is not the benchmark, it is the software and hardware used for the tests, they do not show the results of each test. they just show the final result that is the graph i posted above __________________ So the bullet proof vest aint a $hit when d laser is pointed to your head Kid

05-10-2003, 05:56 AM	#8
Brad Thermophile Join Date: Oct 2001 Location: Nuu Zeeelin Posts: 3,175	they are just using select benchies, just like intel did when the P4 was first released __________________ 2x P3 1100's at 1400, Abit VP6, 2x Corsair 256mb PC150 sticks, 20gb 'cuda ATA-III, 2x 40gb 'cuda ATA-IV in raid 0. 20" Trinitron. No fans 2x 2400+ at 2288mhz (16.0 x 143), Iwill MPX2, 2x Kingmax PC-3200 256mb sticks, 4x 20gb 60gxp in Raid 5 on a Promise SX6000. Asus Ti4200 320/630. Cooled by Water

05-12-2003, 12:13 AM	#9
Rayman2k2 Cooling Savant Join Date: Feb 2003 Location: Annapolis, MD Posts: 190	well, by personal experience, the 3.06Ghz w/ HT ran somewhat faster than w/o HT, especially when the computer froze, and I was able to kill the frozen task with the "virtual" processor. __________________ -Insert Signature Here-

05-12-2003, 06:17 AM	#10
Balinju Cooling Savant Join Date: Dec 2002 Location: Malta Posts: 495	i saw some benchmarkings of the 3.06ht and 3.06 w/o ht. in rendering benchmarks the 3.06ht smoked the 3.06 w/o ht. and also it smoked the 3000+ barton but i still prefer AMD __________________ So the bullet proof vest aint a $hit when d laser is pointed to your head Kid

05-09-2003, 07:20 AM	#2
JSimmons Cooling Savant Join Date: Apr 2003 Location: 4-sided room with an exit going east, and an exit going south Posts: 392	They were running the test with a 16-bit app?

05-09-2003, 07:49 AM	#3
pelle76 Cooling Savant Join Date: Apr 2003 Location: Sweden, Skovde Posts: 101	I wonder if they really have gotten those HyperThreading scheduling algorithms right yet. If they haven't then the lesser result with hyperthreading on doesn't suprise me. I guess HT requires a little more overhead, and therefor without a proper scheduling algorithm the performanceloss is greater than the gain... ???

05-09-2003, 12:25 PM	#5
Blackeagle Thermophile Join Date: Oct 2002 Location: U.S.A = Michigan Posts: 1,243	Balinju, Didn't find any benches there. Mistake? Taken down?

05-12-2003, 12:07 PM	#11
Blackeagle Thermophile Join Date: Oct 2002 Location: U.S.A = Michigan Posts: 1,243	Or is Intell the bigger & better liar??? After reading this tread I did some looking around. I was concerned as I'm thinking on buying a XP3200+ once they are out and the price drops a bit. At the Inquirer I read a article on this issue that make referance to a PCWorld set up benchmark made up of differant apps. that are in real world use, not synthetics (sp). Seemed a good idea to me, after all I'm not into benching. According to PC World the XP3000+ does beat out the P4 3.06 CPU, just not by 17%. The 3000+ scored 137 in their benches, the highest ever scored on the best system tested. They ran the same tests on two other high end set-ups from other suppliers and they scored 134 & 136, close. The p4 3.06 best score on these tests was a 132 (three systems also), while the worest was a 117 with the last system ran them at 127. Comparing scores the 3000+ came out on top, in these tests. And for me the scores on a test that isn't a real app is less impressive than tests run on real apps. ALL the differant top benchies these days seem to show bias. And it seems Intell is by far the big winner in that bias. So for me tests makeing use of real apps are the ones I'll go on. I think a comparison of past benches and present ones show this to be a real problem. But that has been covered by many others better at judging than I already. I know this won't matter to those who love to try to set new high marks in 3DMarks or other benchies. They have to go with what the benchies are, just the way it is.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)