http://www.expreview.com/news/hard/2...8206d6758.html
http://www.expreview.com/img/news/07...3899489_rs.jpg
http://www.expreview.com/img/news/07...6409212_rs.jpg
http://www.expreview.com/img/news/07...6047070_rs.jpg
if repost please delete.
Printable View
the memory BW suggests another 'bad config' case
...a few more weeks till some undeniable and unquestionable benches:)
Shows some promise, just need to ramp the clocks up by 50%
I guess the question on my mind is - can Phenom, supposedly debuting @ 2.4GHz, defeat the 3.2GHz X2 6400+? Looking at these results, if they are indeed valid, would suggest the 3.2GHz X2 would hold an edge over the 2.4GHz Phenom X4 in single/double threaded applications. Only in heavily threaded applications/benchmarks would Phenom come out on top... IMO that would be sort of embarrassing and bad PR if that were to be the case.
Perhaps it would be best to launch Phenom @ 2.6GHz+? Stuff low TDPs, 125W for a quad is perfectly acceptable, at least have your next gen chip OUTPERFORM your last gen chip in all benchmarks, that should be the minimum requirement.
pwnt?
lame
What's embarrassing about that? Hell we have similar scenario today with Q6600 vs. X2 6400 Black!Quote:
Originally Posted by Epsilon84
BTW there's no such thing as quad core GP 7000 Phenom!
Gosh, why turn everything into Intel vs AMD?! I'm simply talking K10 vs K8 here. Due to the low clock rate of Phenom, it will most likely be outperformed by the X2 6400+ and perhaps even the X2 6000+ in non heavily threaded applications.Quote:
Originally Posted by Nedjo
Thus, in certain benchmarks we will have a case of old outperforming the new. That is bad PR.
Unlike AMD, Intel actually has equivalently clocked quads (albeit much more expensive). That is the difference.
My thoughts too.Quote:
Originally Posted by Nedjo
then Phenom GP @ 3GHz would have decent power but I bet even Phenom X4@ 3.2GHz will lose to Yorkfield @ 3.2GHz
I agree with your assessment of the K10. I'm waiting for Phenom before my next build, because I want to go AMD. But not if their new architecture gets outperformed by their old line.Quote:
Originally Posted by Epsilon84
Well, we've already seen 3GHz+ OC for Phenoms, so outperfoming by old series is kinda not possble.
And still nobody have ever seen "famous" B2 revision's results.
Not much time left to wait and see.
I'd wanna see some proper comparisions where CPU performance is measured, not system and VGA performance thanks. :)
That sandra memory benchmark looks like single slot with DDR2-1066 but they used two memory modules.
CPU and multimedia is more than twice as fast as my X2 3800 at 2.0 GHz.
I assume the phenom sytem is more energy efficient (89W TDP) that the x2 6400 system (120W TDP).
I'd be interested in the idle power consumption of the phenom system compared to an X2 at same clock speed.
4 cores against 2 is useless test ...
Fake me thinks.
do all of those tests use all 4 cores?Quote:
Originally Posted by OBR
Very few of those do.Quote:
Originally Posted by ryboto
:confused:
I would gladly take that setup to see if I could fix the obvious problem with the memory bandwidth.
Overall just more FUD imo.
Wow, how surprising! New gen tech being faster than old gen tech? Pleeeease
My oh my , the doubters are still unconvinced...Heh , we'll wait 3 more weeks until their fantasy world will crumble.
In case of sandra's cpu benchmark, it uses four cores, so is it a phenom x4 or a GP-7xxx (tri-core)?
On an 2x2210 system i get 97% better results using 4 in oposite to 2 cores on the sandra 07 cpu benchmark.
Using the 14500 i get on an X2 at 2 GHZ as the basis would mean 28565 with four and 21420 with tri cores.
So if it's a tri-core it is around 40% faster at the same clock. If it's a quad core it is around 6% faster in that benchmark.
Thats how it appears to me also!Quote:
Originally Posted by Sneil
http://www.expreview.com/
That's why. Who's been the biggest source of FUD in the past few weeks?
Baton 3200+ was faster than Winchester 3000+ years ago. Its clearly that fastest old gen CPU is little faster than slowest new gen.
Compare EE netburst Pentiums with e6300. EE will be faster in most benchmarks for sure. But while having both overclocked e6300 will trash EE. Same with K8 vs K7. Same with 6400+ vs GP7000. If first batch Phenoms can do 3Ghz+ on air then its very possible that first retail Phenoms can hit 4Ghz stable with good aircooling.
You do bring up a good point.Quote:
Originally Posted by Kasparz
LOL man you are so full of :banana::banana::banana::banana: it's not even funny. Some of you guys will do ANYTHING to sugarcoat AMD's current predicament, it's rather sad but slightly hilarious nonetheless. :clap:Quote:
Originally Posted by Kasparz
Firstly, an A64 3000+ is CLEARLY superior to an XP 3200+. You want proof?
http://www.xbitlabs.com/articles/cpu...0_6.html#sect0
The A64 3000+ beats the XP 3200+ in 95% of the benchmarks! Many by significant margins as well.
Next, you want to compare a Pentium EE to a LOW END E6300? LOL cmon, how about comparing it to a freaking X6800... and besides, the E6300 isn't any slower the EE anyway.
http://anandtech.com/cpuchipsets/int...spx?i=2795&p=9
You make up 'facts' to try to justify your argument, hoping people are ignorant enough to actually believe the BS you're sprouting.
As for the first retail Phenoms overclocking to 4GHz on air... LOL! Yeah, AMD just decided to scale back their launch speeds to 2.4GHz and you think they're gonna hit 4GHz on air!!! ROFL!!!
Phenom 7xxx are supposed to be the triple cores... the X4 are the 9xxx models...
Actually, he just got debunked and then some. :up:Quote:
Originally Posted by Zytek_Fan
I told Winchester, not 2.0Ghz 3000+ Newcastle. Newcastle was very rare on s939 and mainly used in s754.Quote:
Originally Posted by Epsilon84
If you want to compare it to x6800, then compare x2 6400+ with fastest Phenom, not slowest.
If you would read it clearly, i'm trying to do comparing with fastest old gen vs slowest new gen cpu's. F3 K8 CPU's are doing 3.5Ghz with air. I told somewhere that F3 chips can do 3.5Ghz, they all laughed. Now when AMD dropped prices and more people purchased these cpu's, we are seeing lot of 3.5Ghz aircooled Windsors.
I'm using Intel for main main system, and both laptops if you would like to know.
Umm... you do realise that both Winchester and Newcastle core 3000+s are heaps faster than an XP 3200+. In fact, as you'll see here, http://www.legitreviews.com/article/118/3/ Winchester is faster than Newcastle anyway, so I don't see your point? If anything you just made your ridiculous point even MORE invalid! Talk about digging yourself a deeper hole?
I've already shown that the E6300 is no slower than the PE 965, what are you trying to prove again?
You do realise the fastest Phenom @ launch will be 2.4GHz, and that it will most likely lose to an X2 6400+ in any benchmark that isn't massively multithreaded.
And WTF, what does F3 K8 overclocking have to do with Phenom overclocking?! You're talking about a 90nm process that has been tweaked for close to 2 years to a new 65nm process on a new architecture that is clearly having initial scaling problems.
Geez, if Phenom was overclocking to 4GHz on air you'd think AMD would be more than capable of releasing it a top bin @ 3GHz instead of 2.4GHz, no?
I don't know if you are blinded by faith, or happily living in ignorance. Either way it doesn't look like facts and reality mean anything to you, so I'll just leave you alone in your fantasy world, where the XP 3200+ outperforms the A64 3000+ and where the PE 965 outperforms the E6300... woops! :ROTF: :ROTF: :ROTF:
All right. Stop this meaning less bickering. It's going to lead to ANOTHER Phenom thread closed. Way to go guys.
So correcting FUD and BS is considered meaningless bickering now?Quote:
Originally Posted by Zytek_Fan
You diverted it from correcting FUD and BS to a flame war. If what he says bothers you, put him on your ignore list. Problem solved.Quote:
Originally Posted by Epsilon84
There is just something about blatant BS that irks me. Ignore list? Nah. Then we'll just have an entire thread full of people like you acknowledging crap like that as 'good points'. Some balance is needed... you know, truth vs FUD... it's a delicate balance. ;)Quote:
Originally Posted by Zytek_Fan
What I meant about the "good point" is that early in the life of something, the previous gen can be faster, however slight it may be. Early days of DDR3 vs DDR2 for instance.Quote:
Originally Posted by Epsilon84
I know about separating truth from FUD is very important, but if people choose to believe it, then let them believe it. Like I said ignore it, or put the person on your ignore list if they keep doing it.
Actually, the only time a new uArch was slower than the previous one was the P4 vs P3, and it was heavily ridiculed at the time.
I won't go as far as to say that K10 will be slower than K8 at launch, just that it will be a mixed bag depending on the benchmark. K10 does have higher IPC than K8, but is 2.4GHz enough to defeat a 3.2GHz K8 in non heavily threaded benchmarks? I seriously doubt it.
Please name me any other new uArch launch other than Netburst where the old gen was actually faster than the new gen?!
K8 was faster significantly than K7 at launch, K7 was significantly faster than K6, and K6 was a clear leap over K5.
From Intel, every generation since the 8086 has resulted in increased performance, again with the exception of Netburst of course.
Don't forget that the first A64 based s940 chips were previewed they were running at like 1.4, 1.6 ghz. Hopefully this means that next year they will bring up yields and bring us higher clocking chips.
no not really.Quote:
Originally Posted by Epsilon84
why is it that a 30% clock speed increase is so crazy. Opteron scaled from 1.8Ghz to 3.0Ghz after a bit... Phenom very well might have the same potential... let's not forget 45nm as well...
I'm hoping what they're saying is right. There is no reason NOT to hope.
I'll be blunt though
I bet you were one of those people refreshing newegg like mad for the prescott release...
Epsilon84, your tone is completely inappropriate for discussion here at XS. Please tone it down.
That and baiting/flaming is also against the rules...need I remind you.
Who's going to protect the forum from such BS posts ?Quote:
Originally Posted by Zytek_Fan
Replying in an acid manner to a stupid post isn't well tolerated , but who takes action against the stupid post ?
For the sake of political correctness we're left with BS.Nice way to go.:down:
Noted, the FUD got to me, I apologise. I will try to be more civilised in future. :up:Quote:
Originally Posted by Vapor
given the 60% clockspeed difference... it's not too bad.
Discussion is discussion..If you disagree with a posters point of view then show facts to support your viewpoint..Quote:
Originally Posted by savantu
No one takes action against the stupid post as you put it as it is a persons opinion. What is actionable is someone else flaming the person not disagreeing with him.
Now Vapor asked for some courtesy here. Lets have it.
Thanks for reading.
right, tri-core 2GHz K10 vs. dual-core K8 3.2GHz :DQuote:
Originally Posted by Dragoran
hence K0 has decent results
The original source corrected themselves and calls it a Phenom 9000 now:
http://myevilprocessor.blogspot.com/
if all the cores are not used then these results are great! stunning if only one core is used
I see only SS
http://aycu26.webshots.com/image/322...6409212_rs.jpg
what was the reason they named tested K10 cpu Phenom 7000 instead of 9000?
I think it was really Phenom 7000 - Toliman triple-core
http://www.expreview.com/img/news/071028/score.png
http://www.xtremesystems.org/forums/...d.php?t=163780
..for the intel fans please refer :) :spam:
yes i know this thread is about k10 vs k8.
and to compare 2.0gigk10 with 3.2ghzk8 cpu is hard to do. why dont they put a 2.0GHz k8 next to the 2.0GHz phenom ? ? ? ? ? ? ? ? ? ? :nuts: :confused:
Precisely. Specially with tests that are GPU limited for some.Quote:
Originally Posted by adamsleath
Even in the Crysis test it's only 7.5% slower than the similar clocked Yorkfield, so the stock performance is not that bad. But the Yorkfield will definitely oc higher.
when is phenom coming to desktop anyway?
yorkfield, is it really great compared to the conroe? in ocing i don't think its way yy better. phenom oc potential might just not suckQuote:
Originally Posted by awdrifter
Maybe they thought the model numbers from July was still valid, when the 7000 was referring to a quad-core model.Quote:
Originally Posted by MAS
A little research over CPU test suite in PCMark05. "CPU test suite" hardly depends on memory amount or HDD speed.
Athlon64 X2 2.0GHz: ~4100
2xDual Opteron 2.0GHz: ~5200
Core 2 Duo 2.0GHz: ~5100
Core 2 Duo 2.4GHz: ~5900
Core 2 Quad 2.4GHz: ~7700
Unfortunelly there are no results for 2.0GHz Intel quad core.
http://service.futuremark.com/compare?pcm05=1036253
http://service.futuremark.com/compare?pcm05=998565
http://service.futuremark.com/compare?pcm05=1037427
http://service.futuremark.com/compare?pcm05=1002841
http://service.futuremark.com/compare?pcm05=1104423
Phenom 2.0GHz: 5830
Athlon64 X2 3.2GHz: 6510
phenom @ 3.0 should 'easily' beat k8@ 3.0...by the look of these figures,Quote:
Phenom 2.0GHz: 5830
Athlon64 X2 3.2GHz: 6510
tho' i have no basis for extrapolation.
?
quad core vs dual core, Adam. Can't be calculated accuratly.
-sigh double post
There is result for 2.0GHz dual dual-core opteron which scores 5162. So Phenom is 13% faster per clock then dual dual-core opteron. It correlates well with the previous data about up to 15% IPC improvements over K8.Quote:
Originally Posted by mAJORD
The original source is HEREQuote:
Originally Posted by accord99
Phenom X4 GP-7000(2.0GHz) updated: GP9000
Athlon 64 X2 6400+(3.2GHz)
ATI RD790
GeForce 8600 GTS
Corsair DDR2-1066 1GBx2
And the naming for 4 cores is 9xxx and tri-core 7xxx, so moving it from 7000 to 9000 4-5 days after publishing is not a sign of genuineness.
KTE check out my post in Crysis benchmark thread here:
http://www.xtremesystems.org/forums/...1&postcount=20
I think you will find it very intriguing ;),and i know you like to do some detective work yourself :D(it's about the famous minimum fps in CPU part of the "tests" and the low score for Phenom).
I'm too tired for detective work, but I'm ready for more benches than just the standard Crysis stuff... wink wink Expreview LOL... :rolleyes:Quote:
Originally Posted by informal
Yeah thanks for pointing, I read it earlier. Just too tied about. I'm picking up a C2D this coming weekend or soon after and can run those tests. Would've got Phenom if they had some around to test. :( Those scores are really all over the place, which makes me doubt the bench heavily regardless of the outcome.Quote:
Originally Posted by informal
Is this it? http://www.strategyinformer.com/pc/c...emo/18345.html
How CPU dependent is the bench? Sorry I've not looked into the bench.
Yep that's the one,grab the demo and run the bat file in the bin directory.Quote:
Originally Posted by KTE
BTW,i noticed the benchmark becomes heavily CPU dependent when you force the "physics option" to HIGH(yep,get inside the demo options,set manually what you want,exit and then run the bat file from bin dir.).The CPU usage jumps to healthy 90%.When everything is set to low,my X2 hovers at lowly 60% during the test.And guess what,i also got the 8.86fps at the famous frame 196.
I looked at the "revised" test done by the expreview and now suddenly the Phenom gets 37.6fps as minimum fps in the same CPU test.
dont know if you saw it but cookerjc said they installed the os again, cause they had strange performance issues with the former test.
fake
That Crysis demo is definitely not a quad core bench: http://www.vr-zone.com/articles/Inte...ns/5368-3.html