Anomalies in Anand's conroe's gaming benchmarks ?

[Gary Key]

Originally posted by: OcHungry

400x7=2800mhz would make FX62 just perfect and faster. Did Anand use this setting?

Please suggest a current AM2 board that will do 400HTT?

 

[OcHungry]

Originally posted by: Gary Key

Originally posted by: OcHungry

400x7=2800mhz would make FX62 just perfect and faster. Did Anand use this setting?

Please suggest a current AM2 board that will do 400HTT?

I don?t have an AM2 board, but know the Asus board used for AM2 goes as high as 425mhz.
I can show you plenty of s939 that will do 400mhz and higher, so there is no reason why sAM2 couldn?t.
But question is: Why 325x9 was not used for FX62 to match Conroe's speed?
If Conroe's FSB is @ 1333, that means the cpu to memory was running @ 333mhz, compared to FX62 @ 200mhz. That is 65% increase in memory speed by itself.
Surely you know IMC can improve overall performance if mem speed is increased.
I get Spi of 35.8ns @ 312 1:1 ratio, but I see FX @ 45ns in your test. This tells me FX was not running at its optimal settings.
I understand that conroe is producing better benchmarks (so far ES and by review sites),
But I believe the AMD's side of things were ignored someways in most reviews.
In any event, I thank you for replying and am looking forward seeing further reviews.

 

[Gary Key]

Originally posted by: OcHungry

Originally posted by: Gary Key

Originally posted by: OcHungry

400x7=2800mhz would make FX62 just perfect and faster. Did Anand use this setting?

Please suggest a current AM2 board that will do 400HTT?

1. I don?t have an AM2 board, but know the Asus board used for AM2 goes as high as 425mhz.
2. I can show you plenty of s939 that will do 400mhz and higher, so there is no reason why sAM2 couldn?t.
3. But question is: Why 325x9 was not used for FX62 to match Conroe's speed?
4. If Conroe's FSB is @ 1333, that means the cpu to memory was running @ 333mhz, compared to FX62 @ 200mhz. That is 65% increase in memory speed by itself.
Surely you know IMC can improve overall performance if mem speed is increased.
5. I get Spi of 35.8ns @ 312 1:1 ratio, but I see FX @ 45ns in your test. This tells me FX was not running at its optimal settings.
I understand that conroe is producing better benchmarks (so far ES and by review sites),
6. But I believe the AMD's side of things were ignored someways in most reviews.
In any event, I thank you for replying and am looking forward seeing further reviews.

1. I own the Asus board and several other AM2 boards. Without serious modification they will not go above 400HTT in a stable manner.

2. This holds true with 98% of the S939 boards, only a couple on the market currently will go over 400HTT as shipped. It still does not matter if you decided to run the same settings with Core 2 Duo.

3. Why not use 11x266 to exactly match the 2.93 Conroe? We did and it does not allow the FX62 to surpass the X6800. Core 2 Duo runs on a 1066 bus, not 1333 so your numbers are incorrect. Also, Core 2 Duo has the lowest memory bandwidth of the AM2 or Netburst offerings, this measurement no longer translates into best performance.

4. You will find out shortly that even running a low latency 1066 memory strap on the FX62 will show great Sandra numbers but means very little in actual applications.

5. The FX in our tests were run at the stock HTT speed, the way the chip is shipped from AMD. Each system was setup and run at its standard configuration. I find it amusing that people (intel or amd) always want to compare an overclocked CPU to a stock CPU to try and state one is better than the other.

6. AMD has not been ignored. I doubt you would find a bigger AMD fan over the past few years as Anand or most of the editors on the site. We are first and foremost, performance fans. As such, Intel offers the highest performing CPU family (Core 2 Duo) now (will in three days). That is not a knock against AMD, it is simply a fact. If you are happy with AMD, stay with them, they make an excellent product. If you want to upgrade at this time then Core 2 Duo offers excellent performance and in most applications, leading performance. Everyone wins in my opinion.