Official AM2 Thread!

[avalanche83]

avalanche83:

Since CAS 1.5 may be the same as CAS 2 (for DDR in A64 setting),
and you have been doing the latency vs timing parameter runs using Everest, could you do another test.

Can you do two runs and compare the latency numbers:
1. 200 MHz 2-2-2-6
2. 300 MHz 3-3-3-9

Ok. I ram a couple more and I think my windows is a little corrupt.

200MHz 2-2-2-6 2T @ 57ns


300MHz 3-3-3-9 @ 49ns


Here is an older pic of 200MHz 2-2-2-6 1T @ 49ns. The cpu is a little overclocked here.

 

[hitechjb1]

avalanche83:

200MHz 2-2-2-6 2T @ 57ns
300MHz 3-3-3-9 @ 49ns
Here is an older pic of 200MHz 2-2-2-6 1T @ 49ns. The cpu is a little overclocked here.

Just to make sure, was the 300MHz 3-3-3-9 49 ns at 2T?

If so, it got lower latency than the 200 MHz (49 ns vs 57 ns) is probably due to some other parameters between the two runs were not at a ratio 2:3 and as a results the higher frequency one gave lower memory latency.

 

[avalanche83]

Yes the 300MHz was at 2T. Could anyone else with DDR2 run Everest Latency to see what they are getting?

 

[dropadrop]

http://www.tomshardware.com/2006/02/21/a_look_at_amds_socket_am2_platform/

First tests... Seems a bit dodgy to me though. Why does the cpu-z screenshot show 100mhz frequency for the memory?

 

[caater]

obviously cpu-z don't support AM2 yet..
if mem was really 100mhz, results would have been quite different..
a bit disappointing, but i guess this is an early ES and with that buggy mem controller..

 

[THE JEW (RaVeN)]

obviously cpu-z don't support AM2 yet..
if mem was really 100mhz, results would have been quite different..
a bit disappointing, but i guess this is an early ES and with that buggy mem controller..

Ai, but at least the thing seemed stable. That's more important to me right now. If the thing sub-performed AND was buggy as h3ll.................well, I'd be a bit more inclined to think that AMD was gonna have a few problems these next few months.

 

[avalanche83]

Wow. Tom's hardware shure don't like AMD. Everythings fuzzy and they don't go in depth on how well they overclock or what the FSB is. They just say that there is no improvement. I guess there would be no improvemen in DDR3 either.

The only good thing that they mentioned was that they where 35w for single core and 65w for dual core which is alot better than the current cpu's at 85w.

 

[OC Detective]

This is a Pre rev F release that THG have their hands on. So basically you have an older rev cpu with a dodgy memory controller on buggy mobo that performs close to its counterpart 939 (4800+). The problem with this review is it merely gives a flavour to the capability of what the AM2 will be like when it does provide DDR2 800 support in its June release with rev F (and fixed memory controller!) cpus.
Oh and yes they claim it was a 2.4Ghz cpu which counteracts the divisible by 333 theory....

 

[c627627]

What do you think these frequencies are about, OC Detective?

 

[Gautam]

The HTT is gonna be 333MHz guys...that Tom's ES was outdated and isn't worth jack. Btw that was not DDR2-667, it was DDR2-400 at 4-4-4 timings, big surprise that 2-2-2 DDR-400 outscored it.

 

[OC Detective]

Gautam, that is what seemingly Sentential, c627627 and I (and perhaps quite a few others) are thinking but every single website I have seen that has published frequencies such as the one c627627 has shown, THG et al seems to contradict this!
I even wrote to THG when they showed the claimed frequencies to challenge them on this subject but did not get any reply.
Anyway I still think it will be divisble by 333 FWIW - the only other possible scenario is a HTT at 400MHz!

(c627627 - yes I saw that table a few months ago)

 

[Gautam]

Really?

"Please note that all of our benchmark results were obtained using pre-production processors that did not yet support full DDR2-800 speed. Thus, the results show the basic performance difference between current Socket 939 systems at DDR400 speed and ideal CL2.0-2-2-5 1T timings, versus DDR2-667 at average CL4-4-4-12 timings"

Yeah really. Check out the raw bandwidth. That shouldn't be effected much at all by latencies, you'd have to agree. Those numbers point to DDR2-400 period. I have a feeling Tom's was too dumb to figure out what speeds they were even running themselves. Of course the CPU-Z means nothing, if it did they'd be running at DDR2-200. This isn't the first time Tom's has completely butchered a review. It seems like they can never do anything right. They aren't exactly Intel fanboys since they manage to cripple those whenever they get they chance as well.

And as for why AM2 will be faster, I've got ram that's capable of about 400MHz 2-2-3, 500MHz+ 3-2-4. I keep my tight latencies, and I add about 200MHz in frequency...now if that won't deliver a gain, I don't know what will.

Now the funny thing is I was preaching exactly the same as you back when 939 debuted. Of course the gains won't be earth shattering, but they'll definitely be there.

 

[Gautam]

Gautam, that is what seemingly Sentential, c627627 and I (and perhaps quite a few others) are thinking but every single website I have seen that has published frequencies such as the one c627627 has shown, THG et al seems to contradict this!
I even wrote to THG when they showed the claimed frequencies to challenge them on this subject but did not get any reply.
Anyway I still think it will be divisble by 333 FWIW - the only other possible scenario is a HTT at 400MHz!

(c627627 - yes I saw that table a few months ago)

THG is full of crap...that's all that needs to be said. And I can also see 400MHz being a possibility.

 

[c627627]

...in June 2006? Surely not across the board.

At first I thought for FX-only, but when they started talking about round number frequenciess everywhere you looked even for x000+s, then I thought they'll do DDR II 800 for Brisbane 65nms...


But 06/06/06 release date frequencies, I find it strange that none of the review sites is even talking about how round numbers & DDR II 667 don't fit...

 

[Gautam]

...in June 2006? Surely not across the board.

At first I thought for FX-only, but when they started talking about round number frequenciess everywhere you looked even for x000+s, then I thought they'll do DDR II 800 for Brisbane 65nms...


But 06/06/06 release date frequencies, I find it strange that none of the review sites is even talking about how round numbers & DDR II 667 don't fit...

Well, the HTT is 333MHz as far as I know.

But one thing to keepr in mind about A64's is that the HTT has no bearing on how fast the memory runs. It's entirely possible for AMD to keep a 333MHz base frequency and still run DDR2-800. We could see 333x6=2000MHz, alongside a CPU/5 memory divider. This is something that could just come up in new motherboard, or even simple BIOS revisions, sort of similarly to the changes in HT bus width being dealt with simple changes in HT multipliers. The A64 is incredibly versatile in this respect. You can have round HTT's, and not-so-round memory frequencies.

As far as June 2006 goes I can almost assure you that we will see nothing but 333MHz base HTT frequency. Almost definitely just 333MHz memory frequency as well. But there's plenty of room for easier tweaking than meets the eye.

And Damien, I guess I will have to agree about the bandwidth, since many members here can attest to my response when 939 came out...isn't that right, OC Detective?

But I think AM2 will deliver slightly better gains, simply since there is a loss of latency in introducing a second channel that blunted the full effect of the bandwidth gain.

 

[avalanche83]

c627627, in that chart it says that those cpu's are 89w. In that review it says that they are 65w for dual core and 35w for single core. I wonder if Tom's Hardware thought that they where 65nm?

Edit: Oh. Never mind. They mention that they must keep those watts to keep up with the 65nm Conroe using the 90nm technology, but never mentions them using that on AM2.

 

[RedDragonXXX]

Socket AM2 CPUs three to five per cent faster

WE HAVE seen many AM2 socket CPUs. Most of them are running at 2.4GHZ and are showcased with DDR 2 800 memory. That is the only way to show any kind of performance difference from the DDR 400 based existing 939 CPUs.

The guys that ran some benchmarks on those machines confirmed that you can expect three to five per cent performance increase and that is about it. AMD claims that those engineering samples are the final revision so you should not expect more of the performance incensement. The existing engineering samples are running at 2.4 GHz but they might end up even faster at the launch date.

However, the price of the DDR 2 will drop and will make it more accessible for the general consumers but DDR 800 wont be anywhere close to mainstream. AMD is playing an interesting game but at least it could support DDR 3 when it comes out, so that is something.

source: theinquirer


This is not good at all. Looks like I will be doing Intel Conroe for my next rig.

 

[Rattle]

This is not good at all. Looks like I will be doing Intel Conroe for my next rig.[/QUOTE]


you and me both bro

 

[RedDragonXXX]

It's going to be sad for me to leave AMD

But what can I say, I go where performance is and also to note there are rumors flying around that AM2 will support DDR3.

 

[Rattle]

It's going to be sad for me to leave AMD

But what can I say, I go where performance is and also to note there are rumors flying around that AM2 will support DDR3.

yeah I seen that but I also go where the performance is, if all we seen of conroe is true its going to be revolutionary, low volts, low temps, beating AMD at their own game, LOW SPEED, Kinda sad in a way I love my NF4 rig