AMD Phenom II Review Thread

[Final8ty]

Just reading though this thread & what a joke from the Usual Suspects[great movie]
This a enthusiasts forum.
I don't care about what res the avg joe runs as im not using it.
I don't care about what cpu the avg joe runs as im not using it.
I do care if the avg joe is being scammed or ripped off because we should all care because we are not experts in all areas & are avg joe under those circumstances.

I don't care about AMD/Intel's market share & only that they both exists so i can buy their products and besides price hiking if one goes under i don't give damn as it aint like the CPU/GfX around the world will suddenly DIE with the company it was made from.

[Eastcoasthandle]

If anyone is still interested:
small review done by me: -->LINK<--

Includes apps mostly not used by anyone else. Comparing Phenom 9950 with Phenom II 940BE at exactly same settings!

Thanks for the review

[Shadowmage]

Just reading though this thread & what a joke from the Usual Suspects[great movie]
This a enthusiasts forum.
I don't care about what res the avg joe runs as im not using it.
I don't care about what cpu the avg joe runs as im not using it.
I do care if the avg joe is being scammed or ripped off because we should all care because we are not experts in all areas & are avg joe under those circumstances.

I don't care about AMD/Intel's market share & only that they both exists so i can buy their products and besides price hiking if one goes under i don't give damn as it aint like the CPU/GfX around the world will suddenly DIE with the company it was made from.

That may be true but it helps to understand the position of a company vs. spewing ignorant stuff like "OMG WHY DOESN'T AMD JUST RELEASE A CPU THAT BEATS NEHALEM". Engineering takes time, and it's a wonder that AMD is even coming close to keeping up, considering that Penryn's CPU core size is 40% larger than Shanghai's.

[informal]

That may be true but it helps to understand the position of a company vs. spewing ignorant stuff like "OMG WHY DOESN'T AMD JUST RELEASE A CPU THAT BEATS NEHALEM". Engineering takes time, and it's a wonder that AMD is even coming close to keeping up, considering that Penryn's CPU core size is 40% larger than Shanghai's.

Exactly.I was already talking about the dedicated core logic area in Deneb/Shanghai and Nehalem designs.Nehalem's core takes up 24.4mm2(without L2) while Shanghai's core takes up 15.3mm2.This is approx. 60% difference,a huge one.Four i7 cores stand at 97.6mm2 so for a proper and logical comparison (from the dedicated core logic POV) we would need an Istanbul whose 6 cores will take up approx. 91.3mm2.Still it's a bit lower than Nehalem's total size for cores,but close enough.
And yes,it's easier to add cores than to design super complex SMT like features,especially if you are AMD in the times of global economical distress and with debt company have to pay.So they opted for smaller core sizes and slightly improved uarchitecture that is still a good 10% faster per clock than Agena(in client type of apps,server types gain even more),over a K10 that would be even more complicated to design and probably more late.That's why we will have Bulldozer ,a new design,at later date(supposedly nothing like K8/K10).

I really wonder if they will manage to launch Istanbul in Q3 or Q4 this year.It would be great if they did,50% more cores in same TDP/ACP envelopes will surely be felt,especially in MT applications.

[savantu]

Where do you have the data about core size in mm^2 ?

[Lightman]

Where do you have the data about core size in mm^2 ?

They were mentioned in couple of reviews on the web.
You know, google is your frient (or yahoo if you prefer).

Good place to start is http://www.chip-architect.com/ .... (older products)

PS. It's time to change you bottom quote

[Final8ty]

That may be true but it helps to understand the position of a company vs. spewing ignorant stuff like "OMG WHY DOESN'T AMD JUST RELEASE A CPU THAT BEATS NEHALEM". Engineering takes time, and it's a wonder that AMD is even coming close to keeping up, considering that Penryn's CPU core size is 40% larger than Shanghai's.

Thats fine but should have its own thread as i came here to see the in's & out of Phenom2 & most of what i see is talk about AMD & Intel themselves & possible futures of them & not about the CPU that is before us which is an enthusiast CPU which does not matter what the avg joe thinks of it because they will not be buying chips in the class/price range bracket anyway.

[informal]

Where do you have the data about core size in mm^2 ?

Here you go:
http://chip-architect.com/news/Shanghai_Nehalem.jpg

And some very interesting numbers from previous designs(both AMD and intel):
http://www.chip-architect.com/news/2...us_Images.html

[savantu]

They were mentioned in couple of reviews on the web.
You know, google is your frient (or yahoo if you prefer).

Good place to start is http://www.chip-architect.com/ .... (older products)

PS. It's time to change you bottom quote

Increased core size is also the result of Intel using fewer layers than AMD ( 9 vs 11 ).That also impacts density per mm^2.

[maroon1]

http://enthusiast.hardocp.com/articl...50aHVzaWFzdA==

Did anyone notice that they were using Three GTX 280 OC in Tri-SLI ?

It might be that Intel processors scale better with more GPU power. So why some people here are bashing hardocp when they don't have proof that shows otherwise ?

[FischOderAal]

Interesting is something else in this review:

Deneb: 2 GB DDR2 800
Core2: 4 GB DDR3 1600
i7: 6 GB DDR3 1600

Do you call this a fair comparison? I'm afraid I don't know how much the amount and the clock of the RAM affects the results.

[largon]

Did anyone notice that they were using Three GTX 280 OC in Tri-SLI ?

It might be that Intel processors scale better with more GPU power.

The tri-SLI 640x480 benchmark makes no sense. Sure, it shows how i7 dominates at resolutions that nobody has used since mid '90s, but such a benchmark has no bearing on realworld gaming performance. The problem is, based on such benchmarks some people will draw unfound conclusions that Deneb would be absolutely horrible for gaming and they would i7 get around double the FPS. But the situation would be totally different when realworld game settings and resolutions are used... "Surprisingly different".

But sure, it is a CPU benchmark. Nothing else. About as relevant as SuperPi.

[Rammsteiner]

Interesting is something else in this review:

Deneb: 2 GB DDR2 800
Core2: 4 GB DDR3 1600
i7: 6 GB DDR3 1600

Do you call this a fair comparison? I'm afraid I don't know how much the amount and the clock of the RAM affects the results.

Yeah, that one is taking the piss

There are loads of reviews where the Intel platform has faster RAM, or even clearly OC'd RAM while the AMD system gets just some overhyped value RAM. DDR800 is just stupid unless they use low latency. About 99% of those reviewers have no single clue what type of RAM the IMC favours running at any speed.

Im not pointing at all reviewers being biased, not at all. In some way it makes sense since it's simply a fact that Intel has been used a lot more than AMD the past few years in the enthusiast world, there are more reviews and more guides for Intel as well. So it ain't hard to favour a certain type of RAM to show your Intel rig off, however in a neutral review compared to other types of CPU (be it from AMD or Intel) I think it's the most retarded mistake to use different types of hardware among the systems. Why not use 2x HD4870X2 on one and a 6800GT on the other system and look which motherboard scores higher in FM benchies

However, in this review there are just so many gaps. The DDR800 or DDR1066 problem for Phenom, well, you can live with that. But there's no way you can justify why one system would have 2GB of slow RAM and the other 4GB of super fast RAM. I can understand i7 having more RAM, no problem with that, but that C2 vs K10 thingy, that's just impossible Half of that would be already used for the OS, drivers and apps

[mAJORD]

Anyone who takes the slightest bit of notice of HardOCP's tri-sli results needs to look again..

I don't know why the i7 platform gets such huge FPS compared to anything else, but it's been seen before also.

If you look at the scaling between PhII @ 3.2 and at 3.8, you can see it's barely any different. If it was a CPU (and i mean just CPU) related bottleneck, it would take a 6-10Ghz PHenom II to match a 3.2 i7. Totally unrealistic, There's nothing at a technical level to give bloomfield a 300% adavntage at low res. Not even HT.

Penryn isn't much better.. you can't tell me a tri-channel IMC, and QPI can account for that massive difference. Especially when PhII shares similar technology and it doesn't give it any advantage here..

There's a toms Hardware review with similar odd results somewhere also.

I'm not suggesting anyones making up numbers here, just I would LOVE to find out the cause of this. If I had endless cash i'd certainly configure such systems up and investigate it properly.

[savantu]

Interesting is something else in this review:

Deneb: 2 GB DDR2 800
Core2: 4 GB DDR3 1600
i7: 6 GB DDR3 1600

Do you call this a fair comparison? I'm afraid I don't know how much the amount and the clock of the RAM affects the results.

Deneb with 2 GB of DDR2 800 gets better memory performance than Core 2 with DDR3 1600.There is a difference between 1 and 2 GB , but from 2 to 4 or 6 I doubt it's noticeable , if any.

Most benchmarks , except a few games won't have a problem with the memory size ; it's not like they're loading databases into RAM.

[Miss Banana]

About flawed reviews, reading Kyle Benett's joke of a review just makes me very very very sad.
http://enthusiast.hardocp.com/articl...50aHVzaWFzdA==
How in the world can such an ignorant, incompetent fanboy like he is, be at his position, "informing" people about new hardware?

First he uses 4 and 6 GB for the intel systems, and 2 GB for the AMD system and says this about it:

Yes, they have different amounts of RAM. NONE of our test are dependent on RAM volumes and never come close to getting into a cacheing situation. I thought about ramping the Phenom II up to 4GB, but it would have likely made it a bit slower in the scoring. I used 4GB in the C2 system because that is the same stick we have standardized testing on all the i7 systems. The Corsair 1600 DDR3. Had we gone with 2GB in the Core 2 we would have likely gotten a bit faster scores, at least that has been my experience.

Then, to screw things up even more, he tests phenom with 800 mhz ram instead of 1066 mhz because "he can't get it to work". Obviously he gets 1600 mhz memory to work for all the intel systems.

His review is filled with the most biased stupidity in every form and color possible, and then then when someone on the forum asks whether Kyle has something personal against someone at AMD, kyle explains:

The venom is pointed directly at AMD as a company. I will tell you exactly where it came from. A LOSER PART CALLED PHENOM II. Maybe it will be good for Wal-Mart boxes, but I will not be telling any of my family to buy one. AMD deserves a kick in the teeth for bringing this to the desktop market. Should have just called it Phenom scaling up and shut its mouth instead of contributing to a climate of expectation and again letting us all down. Phenom II, what a joke. "Dragon?" You have to be kidding me.

Dear Kyle, there is a reason why every single website on the net gets better Phenom II scores than you.
It's because every single website on the net is more competent at testing hardware, than you are.
Calling the Phenom II a loser part, after your utter failure of a review, is ironic in a sad way.
The only loser here is you, and the people that take you and your website even slightly serious.



On a brighter note, I think the techreport and anandtech reviews are quite balanced ones, that give a good idea of the product.
Did madshrimps make a review yet?

[savantu]

The tri-SLI 640x480 benchmark makes no sense. Sure, it shows how i7 dominates at resolutions that nobody has used since mid '90s, but such a benchmark has no bearing on realworld gaming performance. The problem is, based on such benchmarks some people will draw unfound conclusions that Deneb would be absolutely horrible for gaming and they would i7 get around double the FPS. But the situation would be totally different when realworld game settings and resolutions are used... "Surprisingly different".

But sure, it is a CPU benchmark. Nothing else. About as relevant as SuperPi.

Your post is what makes no sense whatsoever.Kyle tests all playable resolutions ; you just have to turn the page.

What does that tell us ?

Phenom , Phenom 2 and Core hit a limit with tri SLI ; the video cards can do more ; the CPUs can't.

Core i7 moves the limitation on the video cards ; who wants the latest uber high end graphic cards ( geforce 295 , etc ) needs a Nehalem to make the card shine.

In a sense , the test Kyle performed is more relevant than any other gaming test on other review sites.

[Chad Boga]

Kyle is a joke, he is probably the least credible "big time" review site out there.

Wasn't his initial Phenom review also ludicrous?

[Macadamia]

That's how [H] always cripples to meet their agenda.

They probably did the same with *coughcoughcough* certain pre-4k ATI cards under review.

[Miss Banana]

Your post is what makes no sense whatsoever.Kyle tests all playable resolutions ; you just have to turn the page.

What does that tell us ?

Phenom , Phenom 2 and Core hit a limit with tri SLI ; the video cards can do more ; the CPUs can't.

Core i7 moves the limitation on the video cards ; who wants the latest uber high end graphic cards ( geforce 295 , etc ) needs a Nehalem to make the card shine.

In a sense , the test Kyle performed is more relevant than any other gaming test on other review sites.

Uhm, did you check what hardware he used?
Did you even read my last post?

If we ignore all the test-setup stupidity for a second, even Kyle's craptastic attitude, displayed throughout the entire article is enough to severely question any of the results he shows and conclusions he draws.

[Chad Boga]

When will AMD be releasing a X3 based on Deneb?

If they price this right and release it at 3.0Ghz, it should really give the Intel Duals a run for their money.

[FischOderAal]

Deneb with 2 GB of DDR2 800 gets better memory performance than Core 2 with DDR3 1600.There is a difference between 1 and 2 GB , but from 2 to 4 or 6 I doubt it's noticeable , if any.

It doesn't matter that Deneb has better memory-perf as Core2. What you have to maintain if you do a review is that you make it as fair as possible and that is definately not given in this review. Especially when he compares Deneb to the Core2 system in the end, while the tested Core2-System is far more expensive he claims the Core2-System has the better bang4buck (when paired with DDR2) Why didn't he test a C2D with DDR2 in the first place? That would've been a nice and fair comparison.

Plus, imho reviews have to be as objective as possible as well, and that's another point where Kyle Benett fails.

The venom is pointed directly at AMD as a company. I will tell you exactly where it came from. A LOSER PART CALLED PHENOM II. Maybe it will be good for Wal-Mart boxes, but I will not be telling any of my family to buy one. AMD deserves a kick in the teeth for bringing this to the desktop market. Should have just called it Phenom scaling up and shut its mouth instead of contributing to a climate of expectation and again letting us all down. Phenom II, what a joke. "Dragon?" You have to be kidding me.

Most benchmarks , except a few games won't have a problem with the memory size ; it's not like they're loading databases into RAM.

Might it be possible that Tri-SLI-systems need more memory, or more memory-bandwith? I really don't know, as I already pointed out.

(P.S. The last AMD-system I had was an Athlon XP...)

[Rammsteiner]

Deneb with 2 GB of DDR2 800 gets better memory performance than Core 2 with DDR3 1600.There is a difference between 1 and 2 GB , but from 2 to 4 or 6 I doubt it's noticeable , if any.

Most benchmarks , except a few games won't have a problem with the memory size ; it's not like they're loading databases into RAM.

Oh, Ive been using 2GB past months because of the IMC being a kind of dead, and believe me, you will notice 2GB vs 4GB

Only reason why Deneb still gets a better memory performance is because of the IMC, nothing else. This however does not justify to just use 1) 2Gb and 2) under rated RAM.

Would you bench i7 with 3x 512MB running DDR3 1066 while comparing it to other systems? Im sure it will get more performance than anything else, but it's in no way correct

[Macadamia]

When will AMD be releasing a X3 based on Deneb?

If they price this right and release it at 3.0Ghz, it should really give the Intel Duals a run for their money.

2.8Ghz unlocked est E8400 pricing.


Just need to see the GTA4 benches.

[VoodooProphetII]

I just read this [H] review and it's pretty sad to see the figures he came up with compaired to other sites. The review was bias without a question. I would also like to know why he did not go ahead and used a 4870X2, or say two of them so that there was no need to use a Nvidia chipset mobo. The Phenom II would have shined in the game tests when compaired with a 790GX / FX chipset mobo.