VERY IMPORTANT Athlon 64 MULTIPLIER INFO

[StriderGT]

HALF A64 CPU MULTIPLIERS (8.5x,9.5x etc) DO EXIST AND ARE FUNCTIONAL EVEN IF UNDOCUMENTED BY AMD.
There is no diagnostic application (including CPU-Z, Core Center, etc) at the moment that reports CORRECTLY the MEMORY DIVIDERS (200, 166, 133, etc) that result from the use of HALF CPU MULTIPLIERS. OCA64 is going to become the first one to do such calculations providing THE REAL Operating Memory Frequencies...

- MSI Core Center does not even recognize HALF CPU multipliers, let alone calculate the resulting MEMORY DIVIDERS misreporting everything...
- CPU-Z and ClockGen misreport the MEMORY DIVIDER and as a result they misreport the OPERATING MEMORY FREQUENCY of an A64 running with a HALF CPU MULTIPLIER...
- Although there are HALF CPU MULTIPLIERS there are no HALF MEMORY DIVIDERS.

EXAMPLE:

8.5x CPU multi, 1:1 Memory, 248Mhz FSB=HTT, Memory Freq = 234Mhz (divider 9)
Mem Freq is NOT 248 reported by CPU-Z and Clockgen (divider 8.5)

I am 100% positive about this because my memory maxes at 233-234 (3D mark 2003 crash)...

UPDATE: More Detailed Info on CPU-Z 1.26
CPU-Z misreports the memory speed when you use a half CPU multiplier (8.5x, 7.5x etc) AND the 1:1 ratio (memclock 200)
In this case it reports your memory frequency wrong, because there is no half memory divider(whenever you see such a thing in CPU-Z its a misreport)
I checked half CPU multipliers AND 166, 133, 100 memclock ratios and in these cases everything is ok in CPU-Z.
Current versions of all other diagnostic programs have much more misreport issues!

OCA64 v1.22 is online (link below)

PS This is a repost of my message in another anandtech overclocking thread, due to the importance of the information...

 

[Duvie]

I saw this a week ago!!!

NO one responded in my thread so I thought it must be common knowledge....

 

[StriderGT]

Yes, but this is the first time that we have a conclusive idea of what is going on. Even in your post, you trust CPU-Z memory readings with a half cpu multiplier, which is wrong...
This is why I posted front page, because I think that finally I've reached something really conclusive not speculative, example given etc

 

[Duvie]

Originally posted by: StriderGT
Yes, but this is the first time that we have a conclusive idea of what is going on. Even in your post, you trust CPU-Z memory readings with a half cpu multiplier, which is wrong...
This is why I posted front page, because I think that finally I've reached something really conclusive not speculative, example given etc

Actually read it again...I stated they were a bit off, but for the overall the added bandwidth appeared to be there which told me the multipler was correctly working....

The bios was reporting it right on at post.....

I knew it worked....Pretty conclusive was my own results...hardly speculative!!!

 

[StriderGT]

Originally posted by: Duvie

Actually read it again...I stated they were a bit off, but for the overall the added bandwidth appeared to be there which told me the multipler was correctly working....

The bios was reporting it right on at post.....

I knew it worked....Pretty conclusive was my own results...hardly speculative!!!

I did read it again and:
"I set it to 8.5 x 310 = 2635mhz....set ram to 202.7 or 405ddr with 3:2 (133 divider)...."

Yes, this is correct even if you get it from CPU-Z since 3:2 ratio is correctly implemented when A64 Half multiplier is being used. But:

"actually ram speed should be 3/310 for 103x2=206x2 for 412ddr...nonetheless I have ram scores of 5900's for 400ddr (9x300) so it seems to scale right "

Actually Ram speed should not be calculated this way, as we have covered the subject one thousand times in these threads (there are only integer memory dividers), so what does the 5900 prove? That the memory runs at 412DDR, a speed we are sure is invalid!?!

Finally there is a very cryptic and very difficult to understand parade of benchmarks with half info all the time on what benchmark was used, what settings etc.
Yes you had a solid point, but I've read the text 10 times before I finally figured it out by using a lot of guesswork.

PS1 I do not have the energy nor the intent to argue infinetely whether your post covered the subject of half-cpu A64 multipliers, which I think it did not, but it is really a triviality, I would have posted my findings anyway since I hadn't read your post beforehand...
PS2 Again what exactly was the BIOS reporting? 202,7Mhz mem, I guess not...