i figure 2x 246's (rev E) should be a nice start, they should be cheap enough now that 3 speed grades above now exist.
i figure 2x 246's (rev E) should be a nice start, they should be cheap enough now that 3 speed grades above now exist.
I run a dual opteron with 8 sticks of CMX512RE-3200LL. I know all about what Opteron's require... Perhaps you should read the post above mine... carefully.
Gamingphreek
Lifer
- Mar 31, 2003
- 11,679
- 0
- 81
Perhaps you should have been more specific. 3200LL (Low Latency), you did not have the full number (ie the CMX512RE)
-Kevin
-Kevin
No, I WAS referring to CMX512-3200XLPRO when I said 3200XL. The Abit WN-2S+ will allow you to use plain DDR instead of registered with Opterons. That's why I said that. 
- Jul 29, 2001
- 39,398
- 19
- 81
Looks like AMD starts leaking like crazy at ~2.6 like Intel ~3.4. I noticed this myself in "old" 90nm process. After 2600 Mhz the heat and voltage required jump considerably.
Course I don't believe AMD's numbers. They said 3200 90nm Uses 67W and xbit showed it using 32W load. :::Sigh::: who knows. Maybe we'll just have to see when they come to desktop. It's pretty easy to tell how much heat is being produced by loses when you start overclocking and if good air can handle it or not.
Course I don't believe AMD's numbers. They said 3200 90nm Uses 67W and xbit showed it using 32W load. :::Sigh::: who knows. Maybe we'll just have to see when they come to desktop. It's pretty easy to tell how much heat is being produced by loses when you start overclocking and if good air can handle it or not.
Gamingphreek
Lifer
- Mar 31, 2003
- 11,679
- 0
- 81
Originally posted by: ribbon13
No, I WAS referring to CMX512-3200XLPRO when I said 3200XL. The Abit WN-2S+ will allow you to use plain DDR instead of registered with Opterons. That's why I said that.
No you misunderstood it. You still need Registered memory. However BUFFERED memory is not required. 2 different things
-Kevin
- Jul 29, 2001
- 39,398
- 19
- 81
Originally posted by: Gamingphreek
Originally posted by: ribbon13
No, I WAS referring to CMX512-3200XLPRO when I said 3200XL. The Abit WN-2S+ will allow you to use plain DDR instead of registered with Opterons. That's why I said that.
No you misunderstood it. You still need Registered memory. However BUFFERED memory is not required. 2 different things
-Kevin
no thier not, Registered means it put's but a request in its register as a buffer hence buffered. And no you don't need registered mem for this mobo.
Gamingphreek
Lifer
- Mar 31, 2003
- 11,679
- 0
- 81
Look for a full set on benchmarks and you'll see.
Xeon is just a prescott with more cache,the latest Opteron is pretty much identical to an FX 55 (Opteron has more HT links)
Ed says and he's right that 3 Ghz wont run too cool and will chew up more volts etc, but at 2.4 Ghz, AMD is fairly on par with Intel's 3.8 Ghz and the added cache on the 6 series wont do too much.
Its not like If AMD don?t make 3 Ghz the company will break, unlike when Intel couldn?t or rather didn?t go for 4 Ghz it was seen more as a failure as they needed the extra frequency to keep competitive. The 3 Ghz which we?r all after is just a showboating event which huge pay offs in performance
Before Prescott owners points the finger at AMD for having the same problem, please recognise that they have surpassed Intel?s top desktop offering while have a far more cooler running and efficient chip. The 3 ghz milestone we?re all after is just the icing on the cake, 600 Mhz more on the Atrhlon will just make the 3.8 Ghz P4 look like a Celeron (assuming AMD release a 3 Ghz part under the 4600+ name).
With time between now and when AMD?s dual core desktop appear, its plenty of time to tweak the core, SS,SOI, to achieve better thermals and wattage etc.
Its not like If AMD don?t make 3 Ghz the company will break, unlike when Intel couldn?t or rather didn?t go for 4 Ghz it was seen more as a failure as they needed the extra frequency to keep competitive. The 3 Ghz which we?r all after is just a showboating event which huge pay offs in performance
Before Prescott owners points the finger at AMD for having the same problem, please recognise that they have surpassed Intel?s top desktop offering while have a far more cooler running and efficient chip. The 3 ghz milestone we?re all after is just the icing on the cake, 600 Mhz more on the Atrhlon will just make the 3.8 Ghz P4 look like a Celeron (assuming AMD release a 3 Ghz part under the 4600+ name).
With time between now and when AMD?s dual core desktop appear, its plenty of time to tweak the core, SS,SOI, to achieve better thermals and wattage etc.
there is a good probable explanation given in the "official W8Ns+ thread" at 2cpu.com in the motherboard section of the forums.
One thing the overclockers article didn't discuss in detail was the difference between AMD's TDP and Intel's. When Intel says 120W TDP, they mean under "normal" use (maybe more stressful that normal, but definitely less than max power as defined by voltage X max current). AMD specs theirs with voltage X max current, so their TDP numbers seem to be higher than they would be if AMD used Intel's method.
Therefore, when overclockers is talking about 120W opterons, it is quite possible that the power would still be significantly less than a 120W prescott.
Also, for the A64 line, AMD typically used the max TDP for the fasted chip in the series, and then used that same number on all the lower chips even though they wouldn't use as much power. I think they might be chagning that a bit now, since you see differences with the CnQ numbers and the mobile "type" chips, but anyway, just an interesting tidbit.
Looking forward to these chips, but I will probably end up with a 90nm 3200+ of whatever revision is least expensive in a couple weeks.
It looks like the X800XL may finally start showing up at a reasonable price (saw the Powercolor version for $370 and newegg) and so I'll probably not be able to hold off on a new computer much longer.
-D'oh!
Therefore, when overclockers is talking about 120W opterons, it is quite possible that the power would still be significantly less than a 120W prescott.
Also, for the A64 line, AMD typically used the max TDP for the fasted chip in the series, and then used that same number on all the lower chips even though they wouldn't use as much power. I think they might be chagning that a bit now, since you see differences with the CnQ numbers and the mobile "type" chips, but anyway, just an interesting tidbit.
Looking forward to these chips, but I will probably end up with a 90nm 3200+ of whatever revision is least expensive in a couple weeks.
It looks like the X800XL may finally start showing up at a reasonable price (saw the Powercolor version for $370 and newegg) and so I'll probably not be able to hold off on a new computer much longer.
-D'oh!
LTC8K6 would you elaborate on your thougts, I get the feeling your hinting that Xeons are faster then Opterons all around, and lol im not going to argue but would you add to that or back it up, hey,just for fun
Jeff7181
Lifer
- Aug 21, 2002
- 18,368
- 11
- 81
Well as mentioned, the current Xeons are Prescott based. Because of the design, performance is extremely dependant on keeping the pipeline full, which lots of memory bandwidth helps to achieve. By memory, I don't just mean system RAM... I mean L2 cache, L1 cache, and actual registers. Too slow, or too little memory will create pipeline stalls that effect CPU's with long pipelines much more than CPU's with short pipelines. So increasing the number and to some extent, the size of registers (the fastest form of memory as far as the CPU is concerned) reduces the dependency on the L1 cache, and the L2 cache, and system RAM. So it makes sense that a design that is MORE dependant on these things to perform well will benefit more by improving these things... doesn't it?
Ed has been biased against AMD for a long time (years), in a number of his ranting's. So take what he says with a grain of salt. This may, or may not be the whole story.
LTC8K6
Lifer
- Mar 10, 2004
- 28,520
- 1,576
- 126
Since I hinted that a test with more processors would likely be won by the Opteron, I fail to see the logic in some of the replies to my posts.
I will await more reviwer's explanations of the exact reasons for the Xeon's improvements, as well as more and better reviews. The bandwidth starved argument seems like a good one though. That's why I hinted that more processors would change the results.
Since the 600 chips are out now, we can expect more comprehensive reviews of them soon as well. I hope.
Jeff7181
Lifer
- Aug 21, 2002
- 18,368
- 11
- 81
If you want more evidence of my bandwidth argument, take a P4 2.4B and compare it to a P4 2.4C with Hyper-Threading disabled. The only difference between the two will be that the 2.4C has 50% more memory bandwidth than the 2.4B.
*EDIT* Then realize that a 3.6 GHz processor will likely require even more memory bandwidth than a 2.4 since it moves data that much faster.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 22K
-
News NVIDIA and Intel to Develop AI Infrastructure and Personal Computing Products- Started by poke01
- Replies: 384
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
Facebook
Twitter