A Look At AMDs Socket AM2 Platform

[BaronMatrix]

indeed i do not think people realize when they are ranting about how pc3200 performs the same as ddr2 that all that extra bandwidth is not being used it's getting close but its still not there the only way i think a desktop pc could every use ddr2 bandwidth needs would be with a gaming system a bit down the road with a dual graphics card solution a top notch sound card like x-fi
and a physics card but for the average user and even the casual gamer

WRONG! Everything in the computer is a bottleneck compared to the proccessor, with maybe the exception being the graphics card. The memory is always running at full speed. Any increase in speed will result in an increase in overall systim performance. All the RAM bandwith in the world is not enough to supply the processor. OF COURSE it is all being used. However, with the hardrive being exponentially slower than than both Ram and Proc, the performance increase may not seem like that much. It is there though. It would be impossible for it not to be.


It took me 10 minutes to read this thread. There is a lot of misinformation or maybe disinformation going around. The Inquirer says that HT is going to 1400/2800 which would be a base 280MHz, and now Tom's is saying that they are going to a 333MHz base which would mean 1666/3333. Looking at the divisors of RAM (an 8X multiplier gives 2.6), it is obvious that the same speed rating will have a higher base and lower multiplier. Just like people got better speed when dropping the A64 to a lower multiplier and higher base, the same thing will happen with the AM2.

This time, though the speed difference will, IN MY OPINION, be close to 20% at the same clockspeed, especially with DDR2-800 and the tweaking they applied. People must really hate that AMD took Intel out back and beat the shit out of em.

 

[MadModMike]

First off, The Inquirer is about as reliable as a Celeron Processor. As far as the speed of HyperTransport, Revision 2.0 is spec'd at 1400MHz (2800MHz Effective) as maximum speed, but if you did 333MHz x3 that would get u 999MHz (close enough to 1000MHz) and that would keep the 1GHz (2GHz Effective) HyperTransport. Just my thoughts.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[ltcommander_data]

What I have heard is that The Inquirer reported that AMD was planning on introducing 333MHz base speeds for HT in AM2. It was only supposed to be implemented initially on the FX-62 which would have clocked in at 2.67GHz.

http://www.theinquirer.net/?article=28367

However, given that the speed of the current planned FX-62 is now 2.8GHz it seems more likely that AMD is sticking with a 200MHz base HT speed. The multiplier would be too wierd with a 333MHz base. The extra 133MHz over the original FX-62 will probably make up the difference of the HT loss. In any case, with the TDP already raised to 125W it may have been too hot to implement it now. They have to save some features for 65nm anyways.

And for reza666, the 830D is supposed to compete with the X2 3800+ so it would obviously be inferior to the X2 4200+. The 840D or 940D would be more comparable to the X2 4200+.

 

[Pimp]

indeed i do not think people realize when they are ranting about how pc3200 performs the same as ddr2 that all that extra bandwidth is not being used it's getting close but its still not there the only way i think a desktop pc could every use ddr2 bandwidth needs would be with a gaming system a bit down the road with a dual graphics card solution a top notch sound card like x-fi
and a physics card but for the average user and even the casual gamer

WRONG! Everything in the computer is a bottleneck compared to the proccessor, with maybe the exception being the graphics card. The memory is always running at full speed. Any increase in speed will result in an increase in overall systim performance. All the RAM bandwith in the world is not enough to supply the processor. OF COURSE it is all being used. However, with the hardrive being exponentially slower than than both Ram and Proc, the performance increase may not seem like that much. It is there though. It would be impossible for it not to be.


It took me 10 minutes to read this thread. There is a lot of misinformation or maybe disinformation going around. The Inquirer says that HT is going to 1400/2800 which would be a base 280MHz, and now Tom's is saying that they are going to a 333MHz base which would mean 1666/3333. Looking at the divisors of RAM (an 8X multiplier gives 2.6), it is obvious that the same speed rating will have a higher base and lower multiplier. Just like people got better speed when dropping the A64 to a lower multiplier and higher base, the same thing will happen with the AM2.

This time, though the speed difference will, IN MY OPINION, be close to 20% at the same clockspeed, especially with DDR2-800 and the tweaking they applied. People must really hate that AMD took Intel out back and beat the shit out of em.

So are you agreeing with me?

 

[reza666]

ah ok didn't know that. Thanks for the info.
But really the 840D was not all that much different from 830D on paper just 2 time more expensive.
anyhow AMD rocks 😀

 

[hashv2f16]

First off, The Inquirer is about as reliable as a Celeron Processor. As far as the speed of HyperTransport, Revision 2.0 is spec'd at 1400MHz (2800MHz Effective) as maximum speed, but if you did 333MHz x3 that would get u 999MHz (close enough to 1000MHz) and that would keep the 1GHz (2GHz Effective) HyperTransport. Just my thoughts.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

HEY.. don't diss the celeron man

 

[MadModMike]

You're right, Celeron > The Inquirer.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[FalseInformation]

It took me 10 minutes to read this thread. There is a lot of misinformation or maybe disinformation going around. The Inquirer says that HT is going to 1400/2800 which would be a base 280MHz, and now Tom's is saying that they are going to a 333MHz base which would mean 1666/3333.

Do you have *any* knowledge about Hypertransport ? Why would a HT2800 imply a base of 280 8O :?: . Please have a look at hypertransport.org before replying
Sorry if that sounds harsh, but you stated that there is a "lot of misinformation" in the thread but wrote garbage yourself :roll:

As far as the speed of HyperTransport, Revision 2.0 is spec'd at 1400MHz (2800MHz Effective) as maximum speed, but if you did 333MHz x3 that would get u 999MHz (close enough to 1000MHz) and that would keep the 1GHz (2GHz Effective) HyperTransport. Just my thoughts.

Yes you are right, I thought about the same in the first time. I ditched the idea later, when I red that all hypertransport devices are "chatting" to each other during system startup. Then, the chips negotiate their maximum speed, supported by both, and all that happens @HT reference speed = 200Hz. I dont wanna know what will happen if the AM2 CPU wants to negotiate at 333MHz, and e.g. the nforce4 at 200 Mhz...

To our little, "private" benchmarking topic @2400 Mhz:

I do not care about the benchmarks, you are free to choose anyone you have at hand, if your friend could do some test runs.
I state, that there will be no performance gains, you have to disprove that claim ;-)

Therefore, DDR running at the exact same speed as SDRAM with transmit 2x the data. DDR and DDR2 are referred to by their equivalant(sp.) SDRAM speed. So DDR(2) 400 is actually 200mhz but has the same bandwith as SDRAM operating at 400mhz, which of course does not exist.

Your sine and wave explanations are correct, however, please do some read up on DDR2 technology:

http://www.lostcircuits.com/memory/ddrii/2.shtml
I already posted that link but well, I guess it does not matter, if I do it again
That one is nice, too:
http://www.xbitlabs.com/articles/memory/display/ddr2-ddr.html

In short: SDRAM-100, DDR1-200 and DDR2-400 feature all the same SDRAM-cell clock of 100 MHz, hence also the compareable latencies of e.g. DDR1-400 and DDR2-800.

byebye

FalseInformation

 

[Dybowski]

haha you made me look under my desk and i saw a little beige box with a celeron sticker on it, oh good times.

 

[CPU-Z]

Yes I confirm that CPU-Z output is completely wrong, concerning the CPU type, socket, and all memory information (except for the RAS to CAS delay ... strike of luck 🙂

 

[slvr_phoenix]

The Point of the article is to get us drooling over the pretty little silicon that we are going to buy June 7th to put in our computers June 8th!

Over broken silicon? That so far shows no advantage over the old? What's there to drool about? That's it's got a different socket? :lol:

When something even remotely close to the final silicon comes out (and especially if that makes a better showing than this pre-release sample), then writing a halfway decent article to show that will have a point.

But even then, that's some other article. This article raises so many more questions, about the validity of the testing done in the article and the actual relationship between the sample and what should be in the final product, than it provides answers. So I still don't see there being a point. :roll:

 

[MasterWolverine]

Ram Clock Cycles I believe, but the speed of the RAM speed is dependent upon the CPU frequency divided by the divider. As for being dumb, being dumb and not knowing the answer are 2 very diffrent things. Just ask if someone can explain it, don't short change yourself

It is ram clock cycles, hwoever, the CPU speed is dependent upon the RAM SPEED and the clock multiplier.

Also to settle the earlier confusion about DDR memory. A clock cyle is simply one cyle of power. Frequency only exists in AC power. This Cycle is called a sine wave. In your home, in the US your electriciy has a frequency of 60hz, in Europe it is 50. That is to say that the electriciy changes polarity, goes form positive to negative, 60 or 50 times respectively per second. If you were to graph this it looks like a wave, hence the name sine wave. In a computer, Single Data (rate) Random Acess Memory(SDRAM) sends one signal per FULL sine wave a.k.a. clock cycle. Whereas Double Data Rate (DDR) sends one signal per half cyle. It uses the positive and negative sides of the wave seperately instead of together as one. Therefore, DDR running at the exact same speed as SDRAM with transmit 2x the data. DDR and DDR2 are reffered two by there equivalant(sp.) SDRAM speed. So DDR(2) 400 is actually 200mhz but has the same bandwith as SDRAM operating at 400mhz, which of course does not exist.

I hope that cleared up some confusing, if there was any if not, then I feel kinda stupid for telling you all what you already knew.

Actually, thank you for the information for there might be those reading the post that might not understand the differences and might not understand the discussion that has arrised because of it...and you should never have to apollogize for the dispensation of knowledge, it is the only way we learn.

 

[congo]

and now Tom's is saying that they are going to a 333MHz base which would mean 1666/3333

Yea, this is what I calculated, and would LOVE. However, thats assuming they go with a default 5x multiplier. But Im sure mobo's will have that option even if they dont set as default, and that we can acheive such settings.

My only other hope is that OCZ releases a VX line of DDR2 with insane timings at insane voltages, and DFI releases the board to match it. I would gobble up the ram bandwith, 7.4 isnt nearly enough : )

 

[Half_h_omega]

So let me get this strait, a latency of 2 for DDR 400MHz is equivalent to 2/200e06 seconds or 10ns and a latency of 4 with DDR2 800MHz would be a delay of 20ns?

Is this right? Its been bugging me for ages.

Think I might go and ask in the memory forum actually (as this probably isn't the right place).

 

[Senator]

How bout we all just wait a month or two for the next sample to come out before drawing any conclusions. anything else seems premature.

PS

 

[stephenbrooks]

There was some confusion about the performance with DDR/DDR2 RAM. I think in general DDR2-800 4-4-4-12 is like two copies of DDR-400 2-2-2-6 interleaved in time. So it is the first speed grade that is definitively "better" than DDR-400. It also would seem that a dual-core running off DDR2-800 4-4-4-12 would have the same bandwidth on each core as a single core running the DDR-400 2-2-2-6. Have I got it right?

(And by the same logic DDR3-1600 at 8-8-8-24 timings would be four copies of that DDR-400 interleaved, so would give equivalent bandwidth and latency per processor to a quad core).

 

[stephenbrooks]

LOL.

The discouraging thing I suppose is that whichever of my three types of RAM you hypothetically have, your latency isn't any better, they're just "enough" to keep the cores each getting the same rate as with PC3200. Which is pretty dull really.

 

[stephenbrooks]

intel is listing ddr2 @1000 for their 65nm launch.

Interesting - do you have a link to that? I was afraid we'd be stuck with this 200MHz "base clock" for the RAM indefinitely (if you see what I mean).

As for this actual article, to me it looks like (as others have said), the memory bandwidth benchmark is treating the sticks like DDR2-400, not even 667. Never mind, it's only a sample

 

[MadModMike]

"maybe the hope is to get the settings in bios up higher" - Uhm...What?

"intel is listing DDR2 @1000 for their 65nm launch" - Intel already released 65nm, it's called Presler and Cedar Mill.

"we're only as fast as our biggest bottleneck" - That's your CD Drive, and it runs at 7.2MB/s. Blazing speed, aint it?

Punctuation and spelling is important, there are S P A C E S after comma's and periods.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[MadModMike]

thanx mike the hdd is a bottle neck too! and dont misquote me,please.
oh and if your an english teacher go do that.
the info is here on toms look it up.
I have mis quoted info but hey thats what bieng human is about,making mistakes.yes the base is 200,so sue me.unfortunately i couldnt find the link on the cpu so ill guess i misquoted what i thought id read.Hey mikey come judge my grammar.
i call this piece abstract haiku...mike
dig it! :twisted:

Do all the girls at your school like you? They should, such a rad dude like you. I didn't misquote you, I CTRL+C & CTRL+V right from your post "dude".

"the info is here on toms look it up." - No thanks.

"the base is 200,so sue me." - I assume you mean base clock on CPU's, what does that have to do with anything I said?....maybe, this: ""maybe the hope is to get the settings in bios up higher" - Uhm...What?" but I don't think so.

I'm also not an english teacher and I don't care about your grammar, but I did. I haven't heard the word "haiku" since 7th grade english class, that teacher didn't like me, she was always mean and told me "gonna" isn't a real word, how devisating is that to an 11 year old? It broke my wittle heart! 🙁.

BTW: The HDD isn't a bottleneck when it's 8 of them in a RAID 5 in an Opteron 64 Server running Linux Fedora x64, since it gets 4.5GB/s read times, that's faster than some RAM, garsh!

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[MadModMike]

HAHAHAHAHA! i should space out and forget my details more often,so folks like you show up :lol: your po witto haert is bwoken?lmao.
well goody!
And haiku has been around for centuries girlfriend.
in seventh grade you couldnt even begin to grasp its depth.
ill leave it at that and try some ky,it may help with your dryness.

Uhm...............................What?

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[MadModMike]

No, but you made no sense. You did successfully make a fool out of yourself all over the internet, congrats. Now if you could please use American English and Proper Grammar, please restate what you said so I may understand it w/o needing a PHD in Derr Derr Derrrrrr.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[MadModMike]

uhhh you must be kidding or adhd afflicted.do you not see a reply to your reply?
is it not as condescending as your reply?
yup the whole world has 8 hdds in their pcs in a raid 5 config....not
smells like eliticist round here. must be boring ,bieng so linear.

I must extend my "Uhm......................What?" reply here too.

BTW: I don't have 8 HDD's, nor do anybody around me, we just have over 15 in multiple RAID's, it's a shame though, I shoulda said 16, 15 is an odd number 🙁.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[MadModMike]

I sense jealously 😉. To make you more jealous, I have a Dual 246 Opteron server w/ 4 Raptors in RAID 0, hehehehe.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[leexgx]

--offtopic--
hmm no reply button at the bottem any ways
----
stop flameing as well
----

i dono why users are trying to guess if AM2(K9) will be alot faster or little faster then the older 939(K8) socket he states that this test cant be taken likey as the cpu that was used is not the retail chip (real world thats been stated on DFI forums the AMD64 ram contorler has Full access to the ram unlike P4 so speeds tend to not differ that much 2-4% from an good over clock and high timeings or an underclock and very good timeings you just need to ballance them out {i go for lower timeings my self})

AMD i asume thay did not want to jump on the DDR2 bandwagain untill it got to speeds that would not slow the performance down {same thing goes for the Dual core as well but thay made an statement about that}(when DDR2 came out it was silly timeings but now thay are alot better bye the time it comes to REL we should have even better ram)
we still got 3-4 months yet. AMD will not bring out an slower product and advertise it as an an Super fast cpu (unlike Intel its got an Badge oo its faster now )

DDR2 is alot better Tech then DDR1, there are loads of infomation out there but i say DFI forums is an place to start as thay go into explaing alot on how to OC or just info on how things work + links (dono if i am alowed to link to DFI forums use google {DFIstreets or something})

this is my game drive (RAID 0) so i have no spell checker on here so some words mite not be readable heh