AMD Phenom II 940 "Xtremely" Benchmarked

[JAYDEEJOHN]

I think some people only want to read what they want to see, and then stop. It amazes me people that have no clue as to how AMDs oc have this knowledge of how things work for AMD, when the facts are right under their noses. Im just trying to tell people to expect more, and if that doesnt sit well with Intel owners, thats too bad.

 

[roofus]

Yes full of synthetic benches, But if that is how you want to roll, I think most people agree that deneb has about 5% over agena, If you through in a lot of sythetic memory benches some of which inflate the figures drastically i.e mbench write data 29% etc, Surley you can see the flaw here, I will give you games though, If you where just to count the improvements in games alone which are real world then your point is more valid but surely you agree that real world averages are more important?

i think there is value to weighing out real world and synthetic benchmarks. alot of synthetic benchmarks are designed to make CPU's perform alot of calculations in specific areas that no real world software can do. its about showing potential, not necessarily any direct benefit. (i.e. headroom for future intense apps)

 

[JAYDEEJOHN]

Yes full of synthetic benches, But if that is how you want to roll, I think most people agree that deneb has about 5% over agena, If you through in a lot of sythetic memory benches some of which inflate the figures drastically i.e mbench write data 29% etc, Surley you can see the flaw here, I will give you games though, If you where just to count the improvements in games alone which are real world then your point is more valid but surely you agree that real world averages are more important?

Remember when I said I was a gamer? And youre saying my rebuttal is lame? Just include games then, ok?

 

[gallag]

Just take a look at the same thread you got this chart from and you'll see another one with the NB scaling of PII.

And what improvements did a 200mhz nb bump gain, This is what this whole argument is about so is jdj right? Will a 200mhz bump nett 6%ipc?

 

[JAYDEEJOHN]

So, if you throw out certain ones, its ok, but not others? And when I do it, its wrong, but for you its ok? Let me see, I wonder whos going to see more improvement, all those people seeing only 2% in audio encoding, or all those people that game? I can see it now, it smokes at gaming but my itunes are suffering, I dont want to buy it, yea, and average Joe will decide that way for sure

 

[spathotan]

I think we just need some people up here that actually OWN a P2 940 to do some personal benchmarks for us, maybe even some FRAPS fps snapshots if ingame methods are not an option.

Im getting pretty tired of going through toms and anand and other random sites for this **** when its obvious these sites are biased towards Intel since C2D.

While these chips are good, they are still underwhelming compared to Intels much older Q9000's. No amount of benchmarking or discussion will change that.

 

[JAYDEEJOHN]

I said 4-6%, and am sticking with 4%. DDR3 gives less than 1%, but not too much less, around .7%, thats overall. Im just going on by what Ive read and seen, and by AMDs own calculations, which have been spot on so far, and then theres the 4% also

 

[gallag]

jdj we dont even need the circular arguments, This is all about your prediction that a nb bump of 200mhz will nett 6%ipc, I pointed out that that would be more on average than deneb over agena which on a wide selection of real world benches it is, You proved that with select synthetic benches and games the avg can be higher but that is all by the by, Do you still stand by your prediction that 200mhz nb bump will nett 6% ipc?

 

[dattimr]

By the way, here goes a very interesting review:

http://ixbtlabs.com/articles3/cpu/amd-phenom-2-x4-940-p1.html

Core i7 920 Overall Score: 121.
Phenom II 940 Overall Score: 114.

My opinion is that, while i7 is inevitably the fastest thing around, Phenom II is hardly an embarassment for anyone, except for the die-hard fanboys.

Funny enough, most people "disappointed" on this thread are the ones that haven't "hyped" the chip. Just take a look from the first page to the last, we always had most of the results right from the start. Nobody said it would crush Core 2, nobody said it would be equal to i7. Who is bashing the chip right now is the same people who, in the deepest of their minds, expected it to be jump of a brand new arch. Blame yourselves, not AMD. We had slides comparing it against Q9400 months before the launch.

What Jay has been saying all along (and me too, in some instances) is reflected on every benchmark: it's a very respectable improvement in each and every area of the chip, from performance to power comsumption.

Besides, they will sell much better than the first ones. OEMs will have a good time with them, especially the AM3 versions, which will open lots of possibilities for combinations with mobo/memory. It doesn't matter if even a monkey can raise the FSB from 333 to 400 in the BIOS: most humans won't ever touch it. AMD has something that is good enough, not only for the Joes, but, actually, for almost everyone.

 

[spathotan]

My initial stance on P2 940 was optimistic as you can see by viewing the past pages of this thread. While my stance as sense changed, im only biased towards the obvious data provided that it still does not beat the Q9550. The Q9550 was my personal target for the P2 940 to match or pass, and it didnt do that.

 

[dattimr]

And what improvements did a 200mhz nb bump gain, This is what this whole argument is about so is jdj right? Will a 200mhz bump nett 6%ipc?

On the average, yes. It's on a chart on the same page you got yours. PetNorth did the tests. Too bad I don't have the link right now, but you should have, I guess.

 

[gallag]

On the average, yes. It's on a chart on the same page you got yours. PetNorth did the tests. Too bad I don't have the link right now, but you should have, I guess.

Here is the link, Can you show where a 200mhz nb bump netts 6%ipc? http://www.xtremesystems.org/forums/showthread.php?t=214784

 

[dattimr]

I think we just need some people up here that actually OWN a P2 940 to do some personal benchmarks for us, maybe even some FRAPS fps snapshots if ingame methods are not an option.

Im getting pretty tired of going through toms and anand and other random sites for this **** when its obvious these sites are biased towards Intel since C2D.

While these chips are good, they are still underwhelming compared to Intels much older Q9000's. No amount of benchmarking or discussion will change that.

Well, a friend of mine who works for a very respectable hardware store from Brazil will soon put his hands on a sample and review it at home.

Anyway, I'll get me one in a month or two and, if by then you still wanna see the results of some specific test, feel free to ask me.

 

[gallag]

People seem to be on the defensive to much to get what I am saying, I think deneb is good, It is great bang per buck etc and while the ipc improvements are not amazing the higher clocks help a lot, I am just saying that a 200mhz bump in nb speed will not give 6%ipc and to suggest so is only going to hype it up.

 

[jimmysmitty]

It always amazes me how Intel guys want nothing more than AMD to fail due to fear that they may produce something better than what is in their system. Could you imagine how quiet the Intel community would be if AMD ever takes the lead again. The difference is I would never rub in their faces!!

Not really a fear as much as a look at the facts kinda thing. Although I see your point I think that Deneb was hyped to hell by AMD with their 6GHz obviously cherry picked LN2 OC. If most sites are hitting about 3.7-3.8GHz when AMD easily got 4GHz on air there is something to be said about it.

But at least its not as hyped as Agena was.

This is incredible. 19 pages of the same arguments with no end in sight.

Deneb delivered for AMD users. Not the industry.

Yep. Thats exactally what happened. Its a better upgrade than Agena was at the time and great for AM2 users. But for new builds its hard to justify it.

My initial stance on P2 940 was optimistic as you can see by viewing the past pages of this thread. While my stance as sense changed, im only biased towards the obvious data provided that it still does not beat the Q9550. The Q9550 was my personal target for the P2 940 to match or pass, and it didnt do that.

I was more wanting to see it match or pass Kentsfeild on a clock per clock basis. It came pretty close but overall did not.

But not much we can do but wait. Core i7 has a few revs to come as does P II. But my bet is that the AM3 version wont yeild that much better performance overall. Last time AMD switched memory (DDR to DDR 2) there was no change in performance. But we shall see.

 

[dattimr]

Here is the link, Can you show where a 200mhz nb bump netts 6%ipc? http://www.xtremesystems.org/forums/showthread.php?t=214784

Sometimes you have only to scroll down the page a little. Anyway, I said something wrong. The increase on the NB frequency is of 600 MHz, but I still think it will be a 6% increase *when* coupled with DDR3. Otherwise, the NB at 2400 Mhz (which is doable on any of these chips, just look at the other threads) will result on an average gain of 7%.

 

[spathotan]

Agree on AM3 Jimmy. Its safe to assume it will be better than AM2, but im expecting very little. Within 2%.

 

[JAYDEEJOHN]

Let me restate what I said. Its not just the DDR3, its also the HT Link as well. I said 4-6%, depending, but Ive said Im sticking with what AMD has said, 4%. So, that puts it awfully close to Yorkfield, which is also what Ive said. With better ocing to come, as process matures, therell be no reason to pick 1 over the other, except price

 

[jimmysmitty]

Wait one second there. Since DDR3 has higher latencies which directly affects memory performance for AMDs CPUs wouldn't a performance increase have to have that calculated in as well?

 

[Hellboy]

Wait one second there. Since DDR3 has higher latencies which directly affects memory performance for AMDs CPUs wouldn't a performance increase have to have that calculated in as well?

In one sentence you have just killed every AMD fanbois dream...


Thankyou Jimmy, im gonna have to wear incontinent pads to night cos im going to be pmsl to sleep

 

[JAYDEEJOHN]

jimmy, you cant say theell be the same lag from the cache as there was with the move to 65nm. First of all, its already at 45nm, and theres basically no change, so youll only see the improvements

 

[JAYDEEJOHN]

Think about it, if DDR3 was actually slower, then why make it? Or is AMD and primarily Intel just ripping us off, since Intel has been introducing the new ram segments lately

 

[dattimr]

Wait one second there. Since DDR3 has higher latencies which directly affects memory performance for AMDs CPUs wouldn't a performance increase have to have that calculated in as well?

Yes, but DDR3 is already an improvement over DDR2 with the Core 2 Quads, which aren't that sensitive to memory performance.

DDR2 1066 also has worse latencies, but, overall, it's faster for Phenom. Same should *probably* apply to DDR3.

Also, god knows how they are "tuning" the DDR3 controller. Some say they are having some issues, so that they bothered to launch the AM2 versions.

 

[roofus]

i bet you don't see it until it is faster. i prefer delayed to half-baked or under-preforming.

 

[gallag]

That just shows a ph2 2.4ghz 1.8nb vs 9850 2.4ghz 2.0nb and 3ghzph2 2.4nb vs 9850 3.0ghz 2.4nb, It does not show nb scaling from 1.8 to 2.4 at same clocks, Or am I missing something?

It should show

ph2 3.0ghz 1.8ghznb vs ph2 3.0ghz 2.4ghz nb.