AMD Shanghai/Deneb Review Thread

[Ghostbuster]

I don't want to compare desktop vs server, so Shanghai is out of the question. I want to know if I will have to clock P2 to 4,5GHz to compete with a 4GHz C2Q for example.

My guesstimates would put the PII @ 3GHz to be a bit behind Q9550 @ 2.83GHz. Assuming hypothethical perfect frequency scaling, a PII @ 4.5GHz would be about as fast as QX9650 @ 4.2GHz. Of course, there's no such thing as perfect frequency scaling as memory, latencies, HT and NB speed limits are factors as well. And also depends on the applications you are using..

[STaRGaZeR]

And also depends on the applications you are using..

That's why I don't want to mix server results and scaling with desktop

I want a new system badly, but I'll wait for P2 to hit the streets. Even if I end with Ci7, maybe they will drop the prices a bit to put it as an alternative to P2.

[Donnie27]

Proof or no proof... As I've already suspected, NDA or no NDA (we;ve already seen the 6GHz screenshot)... if these chips can overclock uber well easily, we would be seeing 5GHz+ validations already. Furthermore the CPUZ show its stepping 2 and its not ES, meaning retail ones. It goes to show the difference between the PR demo and real world.

Shanghai numbers are already good indicators, then there was the slides which only compared with Q9300 and Q9400.

The 6GHz LN is find for the speed freak in most of here, that's why we come here. Again, AMD could clear up a lot of things up but hiding doesn't help them, it hurts. Hiding only helps to confirm to some folks AMD is bluffing. IMO, AMD is doing this ALL wrong. The speed crown has no REAL influence on most folks in this market. Might please a few fans.

I think some of you guys are confusing two different arguments. I believe this thing hit just what they said it hit. Most folks I talk to said they did too. I DON'T believe it was a Randomly picked chip=P Neither does 8 out of 10 other Geeks I asked. Guys that's not bashing AMD Only crazy folks believe everything said by AMD or Intel.

I was told something by folks who's got Retail processors. Just as I was told by someone here about i7 before it was even close to launch. Not happy with the binned to death 920's (as an overclocker) and 940 is the one to get. Average overclock with an Ultra-120 was 3.6GHz Max Stable. Didn't beat the i7 at the same speed but not much slower and ran cooler with the same heatsink (that and price has me interested). Wait for better AM3 boards. Price and having to wait for the early adopters to get squeezed first.
His comments like mine, "VAR's and AMD feel they can get more out of AMD fans chompin' at the bit so wait 45 days for the price to fall".

[informal]

As soon as Ramm or anyone else gets his chip,we'll know how it OCs on air and performs .Some amount of chips will be available before the official launch.
As for hiding comment,they are not hiding anything,but like Jazzman said,they are keeping the folks interested by leaking bit by bit during longer period of time.It would wear off if they had shown it all 1 month prior to release. As all can see from Chicago event,the chips are very tolerant to volts and run cool.2 or 3 more events like this one are scheduled until the launch so it will be interesting to see what others will say.

[JumpingJack]

Not exactly true...

Don't be too surprised, Shanghai is a shrink where AMD has gone a long way to rectifying things from Barcelona. Other than larger caches, there is not much new.

From the underground circulations, they improved the cache, improved the memory controller, probably tweaked some buffers and did a much better job on 45 nm than anywone would have expected.

A low errata count is not too surprising, Penryn errata vs Conroe was lower as well...

AMD's first native quad design, from A0 to B3, went up to over 350 errata ... and 99.9% of errata are so sparce and hard to trigger it is not really a big issue. If anything, a high bug count simply is a sign of how complex and difficult it is to design a microprocessor.

Jack

[Donnie27]

As soon as Ramm or anyone else gets his chip,we'll know how it OCs on air and performs .Some amount of chips will be available before the official launch.
As for hiding comment,they are not hiding anything,but like Jazzman said,they are keeping the folks interested by leaking bit by bit during longer period of time.It would wear off if they had shown it all 1 month prior to release. As all can see from Chicago event,the chips are very tolerant to volts and run cool.2 or 3 more events like this one are scheduled until the launch so it will be interesting to see what others will say.

Sure but integrated memory controllers are easy to burn up and are NOT as tolerent of voltage. Ask Ramm about the last one he burnt up? That's something AMD and Intel will have to worry about.

[informal]

Sure but integrated memory controllers are easy to burn up and are NOT as tolerent of voltage. Ask Ramm about the last one he burnt up? That's something AMD and Intel will have to worry about.

That's why NB/L3 runs on separate power plane??
You don't need crazy Vdimm volts in order to OC PhII.You have memory dividers and you use multies to OC the CPU,so the ram stays at the same clock all the time.I thought you knew this already ?
All we know is that PhII ran at 1.9V .That's 1.9V for vcore.Uncore part(NB/L3) ran at much lower volts and clocks using constant NB multi and HTT of 200Mhz.

[LiquidReactor]

Has anyone here come across these benchies yet.

Fud or real?

[Ashraf]

Thread moved and stuck.

[eric66]

The 6GHz LN is find for the speed freak in most of here, that's why we come here. Again, AMD could clear up a lot of things up but hiding doesn't help them, it hurts. Hiding only helps to confirm to some folks AMD is bluffing. IMO, AMD is doing this ALL wrong. The speed crown has no REAL influence on most folks in this market. Might please a few fans.

I think some of you guys are confusing two different arguments. I believe this thing hit just what they said it hit. Most folks I talk to said they did too. I DON'T believe it was a Randomly picked chip=P Neither does 8 out of 10 other Geeks I asked. Guys that's not bashing AMD Only crazy folks believe everything said by AMD or Intel.

I was told something by folks who's got Retail processors. Just as I was told by someone here about i7 before it was even close to launch. Not happy with the binned to death 920's (as an overclocker) and 940 is the one to get. Average overclock with an Ultra-120 was 3.6GHz Max Stable. Didn't beat the i7 at the same speed but not much slower and ran cooler with the same heatsink (that and price has me interested). Wait for better AM3 boards. Price and having to wait for the early adopters to get squeezed first.
His comments like mine, "VAR's and AMD feel they can get more out of AMD fans chompin' at the bit so wait 45 days for the price to fall".

pfft so what they are hiding they didn't talk much about 4800 series either may be its there new strategy how about that

[Ghostbuster]

That's why I don't want to mix server results and scaling with desktop

On the server results, I only look at those "desktop" benches, from there its easy to tell its IPC... the TechReport one has many. Anyway, this is also reflected in ExpReview's slides.

Has anyone here come across these benchies yet.

Fud or real?

We've already seen it.. I would place it as FUD.

[AbelJemka]

Proof or no proof... As I've already suspected, NDA or no NDA (we;ve already seen the 6GHz screenshot)... if these chips can overclock uber well easily, we would be seeing 5GHz+ validations already.

6GHz screenshot with no vcore and no cpu name
Allegation based on nothing.
How many chip owner doesn't violate NDA? Almost all.

[RiverRicer]

Not exactly true...

Oliverda's statement was 'Only twelve known erratas', and you come back with a standard disclaimer statement? Get a clue.

[Jacky]

Just found the thread again, if I miss updates, PM me or feel free to update the OP (mods). As far as I can see there has not been much published that qualifies as a reliable/credible review or preview in the previews week(s)?

[iocedmyself]

http://valid.canardpc.com/show_oc.php?id=462676

well just got the gift of uber shinies, from the same batch of chips that the 4.7 air cooled chip and the 5+ (common we all know it was 6.2 before nda crack down)

After a small scare in which my hardware loving cat (black of course) knocked the chip as it sat in it's blue factory carrier on the ground knocking a fair deal of pins out of whack... got them straightened out last night.

So that is 3.6 @ 1.2125v (stock 3ghz 1.35v) Idle temps are aroud 30c in 23c ambient with a TT duoOrb heatsink load temps are 35c after continous load.

Have taken it past 4.5ghz though more interested in finding max clock at lowest voltage atm.

[mongoled]

http://valid.canardpc.com/show_oc.php?id=459673

well just got the gift of uber shinies, from the same batch of chips that the 4.7 air cooled chip and the 5+ (common we all know it was 6.2 before nda crack down)

After a small scare in which my hardware loving cat (black of course) knocked the chip as it sat in it's blue factory carrier on the ground knocking a fair deal of pins out of whack... got them straightened out last night.

So that is 3.6 @ 1.2125v (stock 3ghz 1.35v) Idle temps are aroud 30c in 23c ambient with a TT duoOrb heatsink load temps are 35c after continous load.

Have taken it past 4.5ghz though more interested in finding max clock at lowest voltage atm.

Hello, are you under NDA, if not would you mind running some benchmarks and publish the results here?

-- EDIT --

Oh and welcome to Xtreme Systems

[SoF]

I would love to show some real results right away...had my hands on it yesterday and if it scales with cold, 5.5gig will be easy. 3.6 @ 1.2125v sounds real good, I didn't test it below 1.375V but >4gig it seemed to need 1.42-1.44V (no ACC yet)...

[G0ldBr1ck]

http://valid.canardpc.com/show_oc.php?id=459673

well just got the gift of uber shinies, from the same batch of chips that the 4.7 air cooled chip and the 5+ (common we all know it was 6.2 before nda crack down)

After a small scare in which my hardware loving cat (black of course) knocked the chip as it sat in it's blue factory carrier on the ground knocking a fair deal of pins out of whack... got them straightened out last night.

So that is 3.6 @ 1.2125v (stock 3ghz 1.35v) Idle temps are aroud 30c in 23c ambient with a TT duoOrb heatsink load temps are 35c after continous load.

Have taken it past 4.5ghz though more interested in finding max clock at lowest voltage atm.

Sounds like a nice chip............hope to hear more info from you

I would love to show some real results right away...had my hands on it yesterday and if it scales with cold, 5.5gig will be easy. 3.6 @ 1.2125v sounds real good, I didn't test it below 1.375V but >4gig it seemed to need 1.42-1.44V (no ACC yet)...

Sounds like they will clock good on SB600 as rumored

[Ben_Brown]

I would love to show some real results right away...had my hands on it yesterday and if it scales with cold, 5.5gig will be easy. 3.6 @ 1.2125v sounds real good, I didn't test it below 1.375V but >4gig it seemed to need 1.42-1.44V (no ACC yet)...

Sounds like a nice chip............hope to hear more info from you


Sounds like they will clock good on SB600 as rumored

This seems like good news. I can break out the old K9A2 Plat since I like that board better for some reason. That would be sweet.

[informal]

http://valid.canardpc.com/show_oc.php?id=459673

well just got the gift of uber shinies, from the same batch of chips that the 4.7 air cooled chip and the 5+ (common we all know it was 6.2 before nda crack down)

After a small scare in which my hardware loving cat (black of course) knocked the chip as it sat in it's blue factory carrier on the ground knocking a fair deal of pins out of whack... got them straightened out last night.

So that is 3.6 @ 1.2125v (stock 3ghz 1.35v) Idle temps are aroud 30c in 23c ambient with a TT duoOrb heatsink load temps are 35c after continous load.

Have taken it past 4.5ghz though more interested in finding max clock at lowest voltage atm.

2000Nhz HT/Northbridge clock?
Did you set the multi for NB manually to get the 2Ghz or it was that way right out of the box with the BIOS you are using?

Good clocks

I would love to show some real results right away...had my hands on it yesterday and if it scales with cold, 5.5gig will be easy. 3.6 @ 1.2125v sounds real good, I didn't test it below 1.375V but >4gig it seemed to need 1.42-1.44V (no ACC yet)...

You know we are waiting for you to take it past 5.5Ghz,you did great with Agena so Deneb should be a piece of cake for you

[SoF]

I am still waiting for some info if I can post some "cutted" screens at least. I can tell you Agena was a lot easier in terms of HTT and RAM with the old bios. I strongly believe even if the chips do run on old bios versions it needs an update before starting to kick ass. HTT and RAM was absolutely no-go during my fist 2-3 hours on that Foxconn board. It performed stable as always but as soon as I tried ACC or 1066 RAM it refused to boot. 3D Performance is something I am thinking about...tested 2x 4870 - really fine for AMD but compared to Intel there is still missing a bit performance. I could see good scaling with higher CPU but to do serious scores you will need a lot more than 5 gig that's for sure

[Oliverda]

I am still waiting for some info if I can post some "cutted" screens at least. I can tell you Agena was a lot easier in terms of HTT and RAM with the old bios. I strongly believe even if the chips do run on old bios versions it needs an update before starting to kick ass. HTT and RAM was absolutely no-go during my fist 2-3 hours on that Foxconn board. It performed stable as always but as soon as I tried ACC or 1066 RAM it refused to boot. 3D Performance is something I am thinking about...tested 2x 4870 - really fine for AMD but compared to Intel there is still missing a bit performance. I could see good scaling with higher CPU but to do serious scores you will need a lot more than 5 gig that's for sure

I think that the Foxconn board has some BIOS related issues. You should try the CPU with M3A79-T Dlx or Giga 790GP-DS4H.

[SoF]

no way AMD is foxconn only for me since some time - my all day sys has a crappy foxconn mainstream board, but still 3.2 gig all day and yes I bought it because I wanted to, hrhr...

[LowRun]

Way to go for a personal attack.

Read what amd said about the ocing event and what chips they used then come back and read my comment... then maybe let it settle a few hours and then post again.

Looks like the same BS after a few days

[Face]

AMD Phenom II X4 940 benchmarks

--------------------------------------------------------------------------------
Here we have some early numbers of the Phenom II X4 model 940 running on a AMD 790GX / SB750 mainboard with 2GB DDR2-1066 and ATI 4850 graphics card.

Super PI 1M took 23 secs to complete.

Performance Test AMD X4 940 :

3DMark Vantage (Performance) - P7334
3DMark 06 Build110 (1920X1200) - 10573
CINEBENCH R10 CPU Benchmark (x CPU) - 9746

http://my.ocworkbench.com/bbs/showth...384#post441384