AMD Shanghai/Deneb Review Thread

[informal]

Why you skipped on the first part of my quote?No argument to counter with ? Please refrain from putting me in the "intel products basher club" since i never done that

As for the 1.9V comment,that's purely enthusiast LN2 thing,i didn't see anyone claim this will be doable on air??Ramm's example is a different one,he had a bad luck and his chip died.We haven't had another example here at XS.
So to conclude,a safe zone is 1.55-16V on air,it's only 18% higher Vcore than stock(1.35V) and the chip is very resilient to high Vcore and runs much cooler than i7 at the same OC(drawing less power in full load too).SOI 45nm is not equal to 45nm hk/mg wrt voltage and heat as will be seen from reviews.

[Caveman787]

cute cat but letting it get so close to your new toy, Id feel really nervous and definattelly shoo it away from it.

informal 16v is a little much lol

[Chad Boga]

there is blurry pic of kitty inspecting the the shiny spiny thing under its curious a$$

Sweet Kitty

[iocedmyself]

Yeah as long as the side is on the case i'm not to nervous about kitty flopping on the warm thing, but yeah i chase him off often enough.

Yeah 1.9v is pretty toasty for anything shy of socket A barton core and even then that was a smidge higher than i liked to go. Especailly when these chips will run 2.8-3ghz at 1.0-1.1v it's nearly double the voltage. Keep in mind that the voltages and TDP were decided before AMD realized just how well the chips performed, and IMO it's likely they've left the chips rated so high to get more of an edge on intel.

The issue with taking the Vcore up that high is the linked CPU-NB voltage, which is responsible for most of my temp problems on the phenom 1 chips. But the max temp for these 45nm chips is 60-65c i believe, so as long as your load temps stay under 50c i'd say voltage is probably alright. Though you may want to go by core temp and not mortherboard temp

Now as i know some were wondering about gaming performance i'll just paste in what i was explaining to my uncle when i was telling him how his little gift was faring

the 9850 would let me run the game in 1920x1200 with rendering options maxed out and draw/detail distance at 15/100 and the ingame benchmark reported those settings used 94% of the video card memory, 88% of all 4 cores on the cpu, 85% of the system memory With Average of 25 Frames per second

That was with cpu overclocked to 3.4ghz and memory speed @ 1066mhz

The phenom 2 lets me run 1920x1200 with rendering options maxed out and draw/detail distance at 32/100 and the ingame benchmark reported those settings used 98% of the video card memory, 64% of all 4 cores on the cpu, 65% of the system memory With Average of 58 Frames per second with cpu at stock speed of 3.0 Ghz and memory speed @ 800mhz

[Caveman787]

25-58 fps nice increase specially comparing a 3.4 agena to a 3.0 deneb.
This is what I wanted to hear if Phenom II can come up with good gaming scores it'll be a winner for mainstream.

What benchmark was that in?

and why do you leave your computer on its side?
Just curious because it seems like a mid tower.

[Zucker2k]

Why you skipped on the first part of my quote?No argument to counter with ? Please refrain from putting me in the "intel products basher club" since i never done that

As for the 1.9V comment,that's purely enthusiast LN2 thing,i didn't see anyone claim this will be doable on air??Ramm's example is a different one,he had a bad luck and his chip died.We haven't had another example here at XS.
So to conclude,a safe zone is 1.55-16V on air,it's only 18% higher Vcore than stock(1.35V) [B]and the chip is very resilient to high Vcore and runs much cooler than i7 at the same OC(drawing less power in full load too).SOI 45nm is not equal to 45nm hk/mg wrt voltage and heat as will be seen from reviews.

You mean, 1.60v. But do you care to share with us on what evidence you base these conclusions? I mean, the chip isn't even out yet, there are no reviews, and what we've seen sofar is largely slideware. Am I missing something?

[iocedmyself]

actually it's a lian-li PCG70 full server tower. Have it on it's side from when i swapped the Deneb back in and i plan to rig up some fans over the NB/SB heatsinks to see if that helps stability.

Yes gaming performance seems to be quite impressive, the CPU is actually fast enough to feed data to the GPU now. As for the benchmark, it's the in game GTA IV benchmark, which aside from being surprisingly detailed seems like a good consistent performance comparison.

Also note that the phenom II aside from being 400mhz slower then agena also had the draw/detail distance more than doubled when attaining double the frame rates. So i'm very impressed with real world performance so far, and am preparing to run crysis/warhead, farcry 2, Call of duty 4/5, Mass effect, Fallout 3, 3Dmark06/Vantage and Pcmark vantage starting with stock speeds. Any other gaming bench requests let me know, i probably have it. So look forward to seeing realworld numbers which i'll post ASAP.

[charged3800z24]

Also note that the phenom II aside from being 400mhz slower then agena also had the draw/detail distance more than doubled when attaining double the frame rates. So i'm very impressed with real world performance so far, and am preparing to run crysis/warhead, farcry 2, Call of duty 4/5, Mass effect, Fallout 3, 3Dmark06/Vantage and Pcmark vantage starting with stock speeds. Any other gaming bench requests let me know, i probably have it. So look forward to seeing realworld numbers which i'll post ASAP.

Can you show any screeny's?

[informal]

You mean, 1.60v. But do you care to share with us on what evidence you base these conclusions? I mean, the chip isn't even out yet, there are no reviews, and what we've seen sofar is largely slideware. Am I missing something?

Poor sarcasm attempt.
The 1.55-16 was obviously a typo,it's clear i meant 1.6V.And the evidence is that 1.6V is 18% over stock.Eighteen percent.Repeat that as long as it takes for you to understand it's not much over default Vcore.
The evidence is numerous end user testimonies ,here at XS,that the chip is cool even OCed and overvolted heavily(with good air).It's cooler than i7 and draws less power in load.
You are missing a lot obviously.

[SoF]

i will go for some clock speed on air this week end - i am not yet allowed to show more but talking without any screens about experiences is something I personally do not like. I showed 1.44V 4gig (it is 1m stable but only showing some loops for now) and will go with max 1.55V normal air and maybe 1.6V cold air outside if I find the time. hope this will clear some things

[Caveman787]

Well aslong as someones feeding us info I'm happy it's alot better then you getting in trouble and the AMD policy beating you down and sending you to guantanimo and confiscating the chip

[justapost]

I don't think idle will win i7 though. They have per-core power gating. D:

Normal 10-25% usage however... should be different.

But i7 needs more juice on the uncore region than phenom on the nb section so I expect a lower consumption even in idle for denebs.

C&Q on the 45nm chips is really much improved, but full features aren't enabled yet, you'll actually be able to specify clock speeds you want to scale down to on the AM3 boards. As well as specify individual clock and power per/core i believe. I'm not sure if i can specify just how low the voltage and TDP has been tested at, but it really does put I7 to shame, for example have 1.0 Vcore @ 2.8 rock solid stable

Griffin allows individual frequencies and voltages per core but the platform comes with three power planes there.
Defining the lowest possible p-state is also possible on K10 platforms but it did not make sense with just two p-states.

Just wanted to give update i'm going to reinstall sandra and test interconct bandwidth and cache speeds now. So again any advice suggestions or even requests lemme know

Thanks would be nice. Estimations based on early shanghai results showed big improvements on the intercore stuff. Would be interesting if it's the same on denebs.

[iocedmyself]

SiSoftware Sandra

Benchmark Results
Inter-Core Bandwidth : 4.45GB/s
Results Interpretation : Higher index values are better.
Inter-Core Latency : 76ns
Results Interpretation : Lower index values are better.

Performance vs. Speed
Inter-Core Bandwidth : 1.51MB/s/MHz
Results Interpretation : Higher index values are better.
Inter-Core Latency : 0.03ns/MHz
Results Interpretation : Lower index values are better.

Detailed Benchmark Results
Processor Affinity : CPU0-CPU2 CPU1-CPU3
Inter-Core Bandwidth @ 2x 8kB : 4.38GB/s
Inter-Core Bandwidth @ 4x 8kB : 4.65GB/s
Inter-Core Bandwidth @ 2x 32kB : 4.87GB/s
Inter-Core Bandwidth @ 4x 32kB : 4.85GB/s
Inter-Core Bandwidth @ 16x 8kB : 4.66GB/s
Inter-Core Bandwidth @ 2x 128kB : 5.03GB/s
Inter-Core Bandwidth @ 4x 128kB : 4.72GB/s
Inter-Core Bandwidth @ 16x 32kB : 4.77GB/s
Inter-Core Bandwidth @ 64x 8kB : 4.40GB/s
Inter-Core Bandwidth @ 16x 128kB : 4.40GB/s
Inter-Core Bandwidth @ 64x 32kB : 4.58GB/s
Inter-Core Bandwidth @ 64x 128kB : 2.66GB/s

Performance Test Status
Run ID : AMD Phenom(tm) II X4 940 Processor (Quad-Core, 3.01GHz, 4x 512kB L2, 6MB L3)
Platform Compliance : Win64 x64
Buffering Used : Yes
NUMA Support : No
SMP (Multi-Processor) Benchmark : No
Total Test Threads : 4
Multi-Core Test : Yes
SMT (Multi-Threaded) Benchmark : No
System Timer : 14.32MHz

Processor
Model : AMD Phenom(tm) II X4 940 Processor
Speed : 3.01GHz
Model Number : 9024
Cores per Processor : 4 Unit(s)
Type : Quad-Core
L2 On-board Cache : 4x 512kB, ECC, Synchronous, Write-Back, 16-way, 64 byte line size
L3 On-board Cache : 6MB, ECC, Synchronous, Write-Back, 48-way, 64 byte line size, 4 threads sharing

Performance Tips
Notice 5008 : To change benchmarks, click Options.
Notice 5004 : Synthetic benchmark. May not tally with 'real-life' performance.
Notice 5006 : Only compare the results with ones obtained using the same version!
Tip 11 : Use the 'Switch Chart Type' button to switch between Detailed and Combined charts.
Tip 2 : Double-click tip or press Enter while a tip is selected for more information about the tip.

[Lightman]

WOW!
That's nice!

For comparison here are my results:
SiSoftware Sandra

Benchmark Results
Inter-Core Bandwidth : 3.67GB/s
Results Interpretation : Higher index values are better.
Inter-Core Latency : 130ns
Results Interpretation : Lower index values are better.

Performance vs. Speed
Inter-Core Bandwidth : 1.39MB/s/MHz
Results Interpretation : Higher index values are better.
Inter-Core Latency : 0.05ns/MHz
Results Interpretation : Lower index values are better.

Performance vs. Power
Processor(s) Power : 157.12W
Inter-Core Bandwidth : 23.89MB/s/W
Results Interpretation : Higher index values are better.
Inter-Core Latency : 0.83ns/W
Results Interpretation : Lower index values are better.

Capacity vs Power
Total Cache Size : 18.01kB/W
Results Interpretation : Higher index values are better.

Detailed Benchmark Results
Processor Affinity : CPU0-CPU3 CPU1-CPU2
2x8kB Blocks Bandwidth : 2.76GB/s
4x8kB Blocks Bandwidth : 2.77GB/s
2x32kB Blocks Bandwidth : 2.78GB/s
4x32kB Blocks Bandwidth : 2.80GB/s
16x8kB Blocks Bandwidth : 2.74GB/s
2x128kB Blocks Bandwidth : 2.78GB/s
4x128kB Blocks Bandwidth : 2.77GB/s
16x32kB Blocks Bandwidth : 2.82GB/s
64x8kB Blocks Bandwidth : 2.75GB/s
16x128kB Blocks Bandwidth : 13.07GB/s
64x32kB Blocks Bandwidth : 9.73GB/s
64x128kB Blocks Bandwidth : 4.73GB/s

Performance Test Status
Run ID : AMD Phenom(tm) 9950 Quad-Core Processor (4C, 2.69GHz, 4x 512kB L2, 2MB L3)
Platform Compliance : x64
Buffering Used : Yes
NUMA Support : No
SMP (Multi-Processor) Benchmark : Yes
Total Test Threads : 4
Multi-Core Test : Yes
Cores per Processor : 4
System Timer : 14.32MHz
Page Size : 4kB
Use Large Memory Pages : No

Processor
Model : AMD Phenom(tm) 9950 Quad-Core Processor
Speed : 2.69GHz
Model Number : 8073
Cores per Processor : 4 Unit(s)
Type : Quad-Core
L2 On-board Cache : 4x 512kB, ECC, Synchronous, Write-Back, 16-way, Exclusive, 64 byte line size
L3 On-board Cache : 2MB, ECC, Synchronous, Write-Back, 32-way, 64 byte line size, 4 threads sharing

Performance Tips
Notice 5008 : To change benchmarks, click Options.
Notice 5004 : Synthetic benchmark. May not tally with 'real-life' performance.
Notice 5006 : Only compare the results with ones obtained using the same version!
Tip 11 : Use the 'Switch Chart Type' button to switch between Detailed and Combined charts.
Warning 5010 : Cannot use Large Memory Pages due to lack of privileges.
Tip 2 : Double-click tip or press Enter while a tip is selected for more information about the tip.


Ignore 2.69GHz reported by Sandra!! Real clocks are as follow:
CPU - 3003MHz
NB - 2277MHz
HTT - 1863MHz (9x207)
RAM - 1102MHz 5-5-5-12-18

[informal]

Intercore latency dropped like a rock,to 76ns from ~130.Intercore BW jumped ~21%.All gains are at the same clock of 3Ghz for both Agena and Deneb(Lightman's and iocedmyself's results)

[cumec]

nice benchmark. can you post some other benchmarks like sisoft arithmetik or multimedia?

[faior]

Just preordered one 940

[Lightman]

Intercore latency dropped like a rock,to 76ns from ~130.Intercore BW jumped ~21%.All gains are at the same clock of 3Ghz for both Agena and Deneb(Lightman's and iocedmyself's results)

Correct!
But bear in mind that I'm running NB at 2277MHz and he probably 1800MHz or 2000MHz!!

[jazzwitherspoon]

Just preordered one 940

Where are people preordering 940s at?

[SoF]

one very quick test with wprime...can't get ACC working yet...need to try other bios

[Ben_Brown]

O what a tease

[SoF]

just throwing some wood into the fire Somebody has to change my title from Xtreme Cruncher to Professor Phenom

[informal]

just throwing some wood into the fire

You tease .I almost thought you posted a whole score,but then saw the blacked out part .
Anyway,3.8Ghz @1.44V on air i presume?Thanks for the results ,it's much appreciated

[charged3800z24]

just throwing some wood into the fire Somebody has to change my title from Xtreme Cruncher to Professor Phenom

well I was like wow..results...then I saw a Black bar covering it LOL...

still thanks for the screen shot...

[zif33rs]

Looks very good SoF..cant wait to see some more.



p.s. update your cpu-z? think there is 1.49