Sign in with
Sign up | Sign in
Your question

A Phenom II Bench by....AMD...

Last response: in CPUs
Share
December 3, 2008 6:50:01 PM

There you go..Taken from Hardware.fr

If the image is not working i'll put the link below.



http://www.hardware.fr/medias/photos_news/00/24/IMG0024...

More about : phenom bench amd

a b à CPUs
December 3, 2008 7:04:13 PM

I couldn't get the link to work. And when I looked around their site all I could find was a really old benchmark they decided to post today.

Edit: Oops, now the picture is working. I thought I saw that one a while ago. Looks too high for Phenom II to me.
December 3, 2008 7:09:42 PM

Well i hope they actually get this high. Don't forget that's against an Q9400. We didn't see anything against the Q9550 or similar. Let's wait.
Related resources
a c 106 à CPUs
a b À AMD
December 3, 2008 7:17:46 PM

Yes, and it says "platform." Without specifics about the systems, this is of very little value.
December 3, 2008 8:03:36 PM

Very impressive. It would lose a lot of its psychological impact, I suspect, if the scale started at 0% rather than 75%. Or included the Q9550 or Q9650.
a b à CPUs
December 3, 2008 8:16:50 PM

I believe thats fake, since when has amd/intel openly compared their new product against the rivals and not their own old one?

performance wise it doesn't matter what it says... I do hope that phenom 2 performs way better than the original ever did, but I call bs on that one
December 3, 2008 9:10:43 PM

I hope that benchmark is right. It gives me hope for the next gen of AMD processors.
December 3, 2008 9:14:17 PM

Hardware.fr links to the source of that info, for those of us who prefer English instead of French:

http://en.expreview.com/2008/12/03/phenom-ii-outperform... (posted 12/3/08)

And a link within that source leads to a sister chart:

http://en.expreview.com/2008/11/26/amd-phenom-ii-bring-... (posted 11/26/08)

So a couple things I noted:

1) This is something AMD released.

2) The Phenom II 940 is 3 GHz; the Q9400 is 2.66 GHz; the Q9300 is 2.5 GHz. All have 6 MB L2, whereas higher-end Yorkfields have 12 MB L2.

3) These benches were typically ones that the original Barcelona lost heavily to the Kentsfield, let alone Yorkfields a few months later.

It is strange that AMD would release a chart admitting they're behind the IPC of half-cache Yorkfields (after you make a little deduction). However, I'm speculating that they will be making a more obvious price argument. If these parts cost between Q9300/9400, then it would be budget-sensible to build systems around Phenom II instead of Q9300/9400s.

The gaming chart at first looks rosy, but then you realize the lack of quantification (is it 45 vs. 60 fps, or 130 vs. 170, or 11 vs. 16?) and the suspicious lack of scaling between Q9300 and 9400 suggest they're hiding some platform/driver details.
December 3, 2008 10:36:32 PM

So, for a 10% increase in clockspeed - if compared to its closest rival (aka 9400) [note: one of the 'half-cached' Yorkfields) - , the average performance gain is something around 7%. Throw in the 'full-cached' models + other usage scenarios - the usual PR BS (since this one must be the 'better' case for Phenom II, if it's the one AMD is shouting about) = Penryn should still beat it across the board, except in some few apps.

Supposing this really came from AMD and using some basic reasoning, that is.

Besides, I don't like the inicial data regarding Deneb's pricing. Just because it's a 3 GHz Quad-Core it doesn't mean it should cost more than a 2.93 GHz one (a.k.a Q9550) that will probably beat it in most things.

That's not to talk about the i7 920 and be flooded with comments about the 'full upgrade cost'.

Now, I take a look at the lackluster AM2+ portfolio, the half-broken SB750 (which is usually included in just 1 or 2 models of each mobo manufacturer [which still tends to look like crap]) and the beautiful GA-EP45-UD3R/P...

Go figure.
December 3, 2008 11:07:53 PM

How about running a benchmark against an i7? No? Thought not.
December 3, 2008 11:41:42 PM

Not sure whos read the Annand article on the 4xxx series gpus, but I find that we may be seeing something very close to this. No, they wont have the fastest,bestest cpu out in Deneb, but like the 4870, it wasnt reaching for top rung, ,just close to it, with a great price. All the fanboys need to step back, and understand just what this may mean. Intel will HAVE to lower its prices to be competitive with AMD, thus, we all win
December 4, 2008 1:06:39 AM

quantumsheep said:
How about running a benchmark against an i7? No? Thought not.

Yea,if they run a benchmark with similar clock i7 say i7 940 Vs PhenomII 940 that'd be interesting,but since that unconfirmed source even implied that Deneb loses to the 6mb l2 cache Yorkfield,I highly doubt such a comparison would even come close with i7's hyperthreading.
AMD die-hard fans now should think twice about their options,but at least they will get something better than Phenom 9950be,which might compete with core 2 with a little more tension than now.
a b à CPUs
December 4, 2008 1:37:03 AM

JAYDEEJOHN said:
Not sure whos read the Annand article on the 4xxx series gpus, but I find that we may be seeing something very close to this. No, they wont have the fastest,bestest cpu out in Deneb, but like the 4870, it wasnt reaching for top rung, ,just close to it, with a great price. All the fanboys need to step back, and understand just what this may mean. Intel will HAVE to lower its prices to be competitive with AMD, thus, we all win


And yet if those numbers are accurate, it'll have a heck of a time even just competing with a Q9550, much less an i7 920, neither of which is exactly a top rung part.

Nevertheless, I'm still hopeful. If they can even pull off a part that's competitive in the $300 class, I'll be quite happy, as right now, they're struggling to compete in the $150ish region.
a c 123 à CPUs
a b À AMD
December 4, 2008 3:56:03 AM

Yay a non clock per clokc "We beat them" comparison straight from AMD. Always can rely on that can't we.

So add in the about 15% OC for the Q9400 then lets test it and see a true clock per clock. Or are we going by price now? Maybe efficiency (energy used)?

It keeps changing and is annoying as hell.

And I doubt they will compare it to eve the Core i7 920. If they were going to they would have put it here. But more than likely they were unable to beat it clock per clock or even with a clock advantage so they decided to go with what makes them look good.

Typical marketing trickery. I am sure Thunderman will be all over this soon enough.......
a b à CPUs
December 4, 2008 4:47:52 AM

quantumsheep said:
How about running a benchmark against an i7? No? Thought not.



lol... they dont need to.

1) i7 is not a budget processor/combo by any means
2) it would destroy it based on the combinations of ram you need to even work it
3) AMD processor is going on budget value

screw all these stupid 'compare it to i7' remarks.
December 4, 2008 5:30:22 AM

^ thats just it. If you want to pay more, and Deneb actually outclocks all the yorkies, whats the point in spending more money? So you can say you have an Intel cpu? IF this turns out to be the case, Intels only going to have to lower their prices to compete. It doesnt matter if clock for clock , one is slightly a lil faster, if that other one doesnt clock as high, and costs more. Lets see how this all pans out
a b à CPUs
December 4, 2008 6:13:35 AM

JAYDEEJOHN said:
^ thats just it. If you want to pay more, and Deneb actually outclocks all the yorkies, whats the point in spending more money? So you can say you have an Intel cpu? IF this turns out to be the case, Intels only going to have to lower their prices to compete. It doesnt matter if clock for clock , one is slightly a lil faster, if that other one doesnt clock as high, and costs more. Lets see how this all pans out



i agree but im being realistic... the phenom 2 is not even going to come close to the core i7 just with the ram capacity/requirements alone. although, i will wait and see what 'really' happens. my guess will be it will compete equally with intels core 2 quads at the same speeds but wont get close to core i7 although it will bring AMD revenue because i highly doubt everyones going to jump over to the core i7 boat just yet.
December 4, 2008 6:35:33 AM

I agree as well. I guess from a gaming perspective, itll be closer than people think, or I hope it is anyways. If they come within 3% of Yorkies clock for clock, who knows?
a b à CPUs
December 4, 2008 7:20:29 AM

sighQ2 said:
Here's some non-inflammatory info. 3 stories about Phenom II.
http://news.softpedia.com/newsTag/Phenom+II+X4

Here's a guide to socket AM3 and Phenom II for noobeez and more.
http://www.amdzone.com/phpbb3/viewtopic.php?f=52&t=1358...

Sneak Peak Phenom II OVERCLOX !!!! - at AMD demo in Austin TX
http://hothardware.com/News/AMD-Sneak-Peeks-Phenom-II-O...



i just read an article and i got this from one of the things posted:
On air, the Phenom II X4 was able to hit around 4GHz at 1.55V with 32ºC temps. With water cooling at 1.6V the new Phenom II X4 broke 4GHz with a 39ºC core temp and in a Vapochill setup we saw 4.4+GHz. The final icing on the cake, so to speak, was the LN2 setup where the Phenom II ran stable at speeds well in excess of 5GHz and booted at speeds in excess of 6GHz.

did they really just say at 1.55vcore and 4ghz speed the phenom 2 runs at 32 Celsius on air? i call shenanigans.
a c 123 à CPUs
a b À AMD
December 4, 2008 7:27:35 AM

^I am betting my PC thats idle temps. And think of this as well. Its 45nm.... 1.55v.... does that sound like a good thing to you? I doubt that chip would live happily for long.

As for comparing it to Core i7, they are coing to have to eventually. Once Lynnfield hits they are going to since it will be below the i920+ price range and the mobos will also be much cheaper.

But the Core i7 i920 itself is not that bad. $300 for a chip that can OC to 3.6GHz+ is pretty damn good. And thats from the stock 2.66GHz too BTW.

Lets not do like last time. I want to see Core i7 vs Phenom II. No excuses. How will I know which one to buy if I can never see them compared to each other????
December 4, 2008 8:34:35 AM

Having a higher vcore than Intel means......nothing. So, Intels process and silicon is exactly the same as AMDs? How do we know how EM may happen on AMDs nbew process? I mean, at what voltages and temps? Or are you just going by Intel, and their HKMG process?

Id like to know where EM is most likely to occur, epecially on AMDs process vs Intels, and where thier strengths are regarding this for each one
December 4, 2008 8:40:44 AM

What may be more telling is, ocing a Intel chip and raising the vCore by 15% would burn out a Intel chip, as thats the same % on the Denebs
a b à CPUs
December 4, 2008 3:36:04 PM

Do we even know what the stock voltage is for Deneb?
December 4, 2008 6:43:47 PM

I'm an AMD fan. Still, I don't trust any pre-release AMD benchies. I'll believe the benchies when I see them at Anandtech, Tom's, Xbit Labs etc. My expectation is still that Phenom II will lose to i7 (but maybe not in games). It will lose to newer Core 2 quads in games, but not by much, and it will match Q6600.

At any rate, we will see in January.
December 4, 2008 11:37:22 PM

does it raise any suspicion at all that these are approximately a month from release and the only benchmarks are from AMD themselves??
either they have gold or it is the same old Phenom with modest gains and they are fearing the backlash from their faithful followers that are expecting something special this time.
by now i really expected to see some samples sent to review sites for independent tests. i am not in love with tests conducted with engineering samples but i cant help but wonder WHY they aren't sending these out to let the review sites give them some love.
December 4, 2008 11:38:59 PM

lol nevermind..yip jumped in there with the same sentiment.
December 5, 2008 12:38:26 AM

Except all the sites yipsi mentioned are pro-spintel biased - read biased kinda like spinners who want to bet on some impossible scenario. The bet would cost you.

It's a little while yet b4 the release in Jan 09 and retail will be into Feb 09

This is just hype b4 the release - pre-release info for press. But it's a serious improve over previous and some think Phenom II it will hang a 4870noose on the TLB plagued i7 - that's my bet. And i7's TDP is too huge. The new Phenom II has reduced pow consume also. Sit back and watch :) 
a b à CPUs
December 5, 2008 2:27:10 AM

First off, I don't think the i7 really has much of a TLB bug (if at all, I thought Intel clarified and said it was fine). At any rate, all the chips out now are just engineering samples (since the real chip just started production a few days ago, or so says Fudzilla). It seems the timeline in the past was the real stuff should start heading out soon so the reviews can be written (but still under NDA). If any semi-credible reviews show up before the NDA, it probably won't be for a little while still. But who knows, all this NDA secrecy is so annoying.
December 5, 2008 2:28:25 AM

sighQ2 said:
Except all the sites yipsi mentioned are pro-spintel biased - read biased kinda like spinners who want to bet on some impossible scenario. The bet would cost you.

It's a little while yet b4 the release in Jan 09 and retail will be into Feb 09

This is just hype b4 the release - pre-release info for press. But it's a serious improve over previous and some think Phenom II it will hang a 4870noose on the TLB plagued i7 - that's my bet. And i7's TDP is too huge. The new Phenom II has reduced pow consume also. Sit back and watch :) 



....Are you implying AMD wont have TLB.....Cause then that be amazing i mean. no TLB i mean that's kinda bad right......that would suck for AMD phenom...
a b à CPUs
December 5, 2008 2:29:07 AM

Haha, good one :lol: 
December 5, 2008 3:43:43 AM

EXT64 said:
Do we even know what the stock voltage is for Deneb?



Ive heard its 1.35, tho Ive seen a cpuz shot at 1.32, running just slightly over 3Ghz. Was reading another forum, and according to someone there, the oc's AMD did, and got to 6.2Ghz, they ONLY used the multi. Im sure you can find the info, its out there
a b à CPUs
December 5, 2008 3:46:33 AM

^ I'd rather wait for the official voltages from AMD's site.
December 5, 2008 4:00:58 AM

Also, getting to 3.9, they only used the multi, and couldnt get higher, but when someone else tweaked everything, it went over 4, but its NDA'd, and thats on air
a c 123 à CPUs
a b À AMD
December 5, 2008 4:38:40 AM

EXT64 said:
First off, I don't think the i7 really has much of a TLB bug (if at all, I thought Intel clarified and said it was fine). At any rate, all the chips out now are just engineering samples (since the real chip just started production a few days ago, or so says Fudzilla). It seems the timeline in the past was the real stuff should start heading out soon so the reviews can be written (but still under NDA). If any semi-credible reviews show up before the NDA, it probably won't be for a little while still. But who knows, all this NDA secrecy is so annoying.


Nah he doesn't trust the Intel press report. He even made another post about it and the site even quoted the press release.

Intel stated there was a TLB bug in Core 2. But there is none in Core i7. And any that was harmful to functionality was all fixed by BIOS update before Nehalems release.

And calling THG a spintel site is just stupid. From what I have seen they have been fair on both sides. They test the chips and post their results and say whats good or bad. Hell they even stated that Pehryn was only slightly better than Conroe, which was true unless you specifically used the more enhanced parts such as SSE4.1.

But meh. Think that way if you want.
December 5, 2008 4:48:11 AM

Theres been erata fixes as usual on i7, as there has been on any cpu made. Thimg is, if i7 runs this well crippled due to some kind of stunting thru implementing this "fix", it musta been killer without it
December 5, 2008 5:41:06 AM

werxen said:
.... and booted at speeds in excess of 6GHz.




Or maybe even a round or 2 of crysis.
a b à CPUs
December 5, 2008 6:17:51 AM

The report I read from the lets call it "the in house test at Austin", was they got to 6.2MHz post, but could enter the OP-Sys. If I could Remember which forum I seen it in I'd post a link. (but most of you have aready seen it anyway).
December 5, 2008 10:50:36 AM

perez8434 said:
....Are you implying AMD wont have TLB.....Cause then that be amazing i mean. no TLB i mean that's kinda bad right......that would suck for AMD phenom...


You should know this already from last spring or early summer.

AMD killed the whole series of the first Phenoms model numbered as 9x00

and then

released newly branded models with no TLB problem = models numbered as 9x50

AMD specifically stated they changed the model numbers so customers would kow they were not buying TLB-bug. really old news!!!!

Some members on this site are spreading the old news about the original KILLED line that had the bug. Even posting comparison grafx to show inferior performance using the original 9x00 models - these are not sold anymore. it's old news.
-------

AMD will not allow that to happen again. The NEW Phenom II is to release in January 09 and all indicators are very positive - there are no tests available yet - it's not available RETAIL until Feb 09. But the release will be at the CES show in January.

14 chips coming between January and June 09 - all based on the NEW NEW stuff = Deneb series, based on server chip Shanghai.
called Phenom II and numbered like 940 !!!! :) 

quads
triples
duals

It seems AMDzone site people have figured out most of the details in theory and some think it might be a threat to upper model quads of spintel but there is no formal test, it's still NDA. They have roadmaps and charts and a lot of future info, but no formal info.

Early official release will be for existing AM2+ SOCKET mobos to provide easy upgrade path with a mere BIOS UPDATE.
Most mobo makers have already released BIOS updates - just goto your website re mobo and check your model for the BIOS. There's also an GUIDE about Socket AM3 at the Zone site. you can google if you want.

Most info on Tom's site about AMD is flamery and very negative spin. And very outdated referring to the era of the TLB problem - which was killed long ago. Deneb has to be clean, has to perform, has to overclox, has to be priced well and be competitive.

I experience a lot of opposition from some regulars here who seem (to me) to be pro-spintel and opposed to the truth about AMD. Also a lot of bench tests are biased cos they are compiled with spintel compilers, or they are cherrypik benches to favor spintel. You need to be a programmer to prove it tho - or talk to one. Programmers use a lot of different compilers, as do companies where performance fast is everything - they use their own custom compilations. here are many ways to cheat or twist bench results. You have to watch carefully.
December 5, 2008 11:29:50 AM

Yeh, ok.

Nice to actually see a CPU-z.

He used a cooler from the old chip I guess?? The 9950 was the top Phenom I. But the 940 is a Phenom II.

This oclox result on air is reported on lots of sites but this the first I have seen a cpu-z.

The results and temps are in black in the pix - that is the AMD AOD = AMD Overdrive Utility which includes ACC (Advanced Clock Calibration)

NOBODY knows how it works! Everybody says it makes it too EZ to OCLOX.
but
You must have a mobo with the SB750 southbridge to use ACC (free download) SB750 southbridge is common on mobos with 790GX, or 790FX northbridge
AVAILABLE FROM = (Gigabyte, Asus, DFI, MSI and others too.)

The top performer enthusiast mobo is the one with 790FX/SB750 chipset.
BUT new chipsets are also expected eventually to support socket AM3.

The initial release will be for AM2+ mobos, so customers have easy upgrade path. (2 chips, one is a Black Edition = unlocked!) @3.0ghz and the other is locked @2.8

Then it switches to SOCKET AM3 RELEASE. with 12 more chips releasing from Feb to June 09.

There will be new DDR3-1333 mobos - oclockers recommend use ddr3-1600 for oclox headroom.

It is not known yet - but mobos might be (some or all??) able to support DD3 OR DD2 ram - see mobo makers - the Deneb is believed to have dual IMC = Integrated Memory Controller - there is no definitive info on this yet, but it is probable, at least, and might be a mobo makers option. (?)

Previous HSF are expected to fit AM3 socket.
December 5, 2008 11:45:05 AM

AM3 has both capabilties, while AM2+ only DDR2
December 5, 2008 11:52:45 AM

I am a fan of AMD but I don't work for AMD.

The info I have is from various sites on the net and is the result of hunting it down for recent months. I should get paid for this, but I don't. I have been researching this because I plan on a new system for me - and I want something good because I deserve the best. I am self-taught for way too long thru my own interest in pc hardware. I am a gamer. And I wanna play Crysis, CoDuty, and other FPS shooters - and have an immersive fantasy experience. I have been out of the loop in recent years but have been playing catchup for a while. I don't work in IT either. I would like to :) 

I simply can't support monopoly or ethics related to abuse of monopoly. I prefer to simply talk truth about whatever hardware; but I don't tolerate abuse and flamewars, cos I have been there, done that, and see no use in it. I think it hurts me and you, and deprives us of the truth. I can be nasty; but I prefer peace and genuine communication.

That's all I got for now. I just hope this helps people who would rather make a smarter choice :) 

Peace.

gotta go,

sigh
.
December 5, 2008 11:59:45 AM

thanx to jaydeejohn for that cpu-z link

that's a first for me

[edit] also thanx to EXT64
for his cpu-z post.
It is unusual for Fudzilla to support a lot of what he says :)  but he sure says some interesting things :) 

later

sigh
.
December 5, 2008 12:37:09 PM

unclefester said:
The report I read from the lets call it "the in house test at Austin", was they got to 6.2MHz post, but could enter the OP-Sys. If I could Remember which forum I seen it in I'd post a link. (but most of you have aready seen it anyway).


6.2 MHZ... 8086 power!!
December 5, 2008 3:08:46 PM

Don't just take Fuad's word for it:

Quote:
I wasn't around the LN2 table the *whole* time, and while I don't recall exactly how stable it was at this speed the AMD guy said they did a couple passes on Crysis at this speed. We did a lot of tinkering at different speeds and it was so long ago it all is kind of a jumble what was what...


http://forums.extremeoverclocking.com/showpost.php?p=32...
a b à CPUs
December 5, 2008 11:40:44 PM

sighQ2 said:
I am a fan of AMD but I don't work for AMD.

The info I have is from various sites on the net and is the result of hunting it down for recent months. I should get paid for this, but I don't. I have been researching this because I plan on a new system for me - and I want something good because I deserve the best. I am self-taught for way too long thru my own interest in pc hardware. I am a gamer. And I wanna play Crysis, CoDuty, and other FPS shooters - and have an immersive fantasy experience. I have been out of the loop in recent years but have been playing catchup for a while. I don't work in IT either. I would like to :) 

I simply can't support monopoly or ethics related to abuse of monopoly. I prefer to simply talk truth about whatever hardware; but I don't tolerate abuse and flamewars, cos I have been there, done that, and see no use in it. I think it hurts me and you, and deprives us of the truth. I can be nasty; but I prefer peace and genuine communication.

That's all I got for now. I just hope this helps people who would rather make a smarter choice :) 

Peace.

gotta go,

sigh
.


Wait for 45nm Phenom 2s then =)

They look pretty good

i7 isn't too much of an improvement on Core 2... Let's see what Phenom 2 will be like
a c 123 à CPUs
a b À AMD
December 6, 2008 6:34:48 AM

amdfangirl said:
Wait for 45nm Phenom 2s then =)

They look pretty good

i7 isn't too much of an improvement on Core 2... Let's see what Phenom 2 will be like


I'm going to put my prediction on this. I am willing to be that Phenom II is going to give the same results as Core i7 did over Core 2. Very little single threaded performance increases but nice multithreaded performance increases. While I doubt the increases will be near what Core i7 is capable of (50% in some cases) I know it will be decent.

But I highly doubt they will get more than 10% in ST apps if not less. ST has hit a ceiling and unless they find a way to do the 6GHz on air then we wont see any amazing performance increases in ST from either side, only in MT apps.

And as for the smarter choice, not trying to be mean but the Smarter choice is the one that performs the best for that price bracket not a certain brand.
December 6, 2008 7:23:19 AM

sighQ2 said:


That's all I got for now. I just hope this helps people who would rather make a smarter choice :) 

Peace.

gotta go,

sigh
.


Subliminal message?
!