AMD A64 4000+ San Diego

infinitevalence · Post by **infinitevalence** » Wed May 11, 2005 9:57 am

No need to appologise, any information you post is usefull for people later on who are looking at buying these chips.

Bwall · Post by **Bwall** » Thu May 12, 2005 8:52 am

Ok..really weird stuff going on. On my hunch last night I started playing with memory speed and timings to see how much it affects CPU clock speed. Dropping the memory speed and timings down decreases the amount of CPU voltage needed to boot into windows.
With my UTT's running 1:1 It would take 1.6v to boot at 2800MHz and be anything resembling stable. If I used the 180 divider it took a little less voltage and dropping the timings dropped the required voltage just a little more.
So, following this line I installed my Mushkin TCCD's again since at higher FSB they are much looser timings. What I've found is that something is hyper sensetive to the memory.

Again using the combo of high HTT, memory dividers, and loosened timings kept dropping the volts needed to get a succesful boot into windows. Eventually at 2.5-4-3-8 /166 divider with the TCCD's I got it to boot at 311x9 (2800)@ 1.45v!! It wasn't stable enough to complete many tests but very interesting. I was able to run it at 1.50v and complete all the benchmarks I could throw at it...I'll see if I can get some screens tonight. This was not even possible with my UTT's as I mentioned above.

Maybe I'll take her top off tonight....opinions?

Bwall · Post by **Bwall** » Thu May 12, 2005 9:34 am

eva2000 wrote:thanks for the info maybe i need to get my hands on the 427-3/509-3 latter bios heh

seems i am unable to achieve what OPB achieved with this cpu (2850mhz at 1.45v air), but mine handles more vcore fine so far

Any luck yet?

infinitevalence · Post by **infinitevalence** » Thu May 12, 2005 9:55 am

It sounds like your memory controler is whats limiting you, this is not that unusual as i have seen winchesters with the same problems. if your getting better clocks with the TCCD i imagen its because of the lower voltage requirements of the memory. i guess at this point i would see if your max on UTT or max on TCCD is higher and then figure where you can take it from there.

Bwall · Post by **Bwall** » Thu May 12, 2005 11:46 am

I don't think it's really held back per say, it's just a really strange beast....
With TCCD I can hardly get it running default settings of 12*200 2-2-2-7 or even 2.5-3-3-7. What's really strange is I went from booting and benching 311*9 1.5v to failing to boot at 311*7 1.45v ...yet 1.45v will still boot 311*9 all with identical memory timings and divider. TCCD takes TONS of tweaking to get things right.
edit: Forgot to mention that with TCCD I can't get anything to boot with a 10x multi. Yet it ran 290x10 on Nate's DFI SLI board. I really need to try a new BIOS.

With UTT in it will boot anywhere, any multi as long as there is enough voltage to support it.

Once I settle on a clockspeed I'll get extensive testing on what's 24/7 stable with loose timings and tight timings. They may end up being the same but right now my testing tells me that with looser settings, I can boot windows with less volts.

eva2000 · Post by **eva2000** » Fri May 13, 2005 2:04 am

wickedld9 wrote:
eva2000 wrote:thanks for the info maybe i need to get my hands on the 427-3/509-3 latter bios heh

seems i am unable to achieve what OPB achieved with this cpu (2850mhz at 1.45v air), but mine handles more vcore fine so far
Any luck yet?

not getting past 2808-2833mhz at 1.67v-1.73v it seems for superpi/benchable

but mem controller seems better than my winnies for TCCD at least got an extra 5-10mhz out of my Gskill PC4400LE

not bad at stock cpu speed of 2400mhz

@8x 305mhz 2.5-4-3-7 1T

eva2000 · Post by **eva2000** » Thu May 19, 2005 9:57 am

this is where i am with BH-5 ram so far @2750mhz at 1.62v

System
4000+ CABHE 0515 SPMW water cooled
DFI NF4 SLI-D 414-3 beta bios
2x eVGA 6800U PCI-E SLI 72.03 forceware drivers
2x 256MB KHX3200K2/512 BH-5
120GB WD 1200JB
650W Silverstone SLI PSU

WinXP Pro SP2
NF4 6.53 remix chipset drivers

- note SLI mode will probably reduce cpu related performance slightly so take that into a/c with the benchmark results.

2x 256MB KHX3200K2/512 BH-5

@10x 275mhz 2-2-2-6 7-17-2-2-1-2 at 3.7v

VID = 1.25v
VID Special = 33%
chipset = 1.6v
LDT = 1.5v
LDT 3x

CPC = Enabled
CAS Latency(CL) = 2
RAS to CAS(Trcd) = 2
Min RAS# Active time(Tras) = 6
Row Precharge Time(Trp) = 2
Row Cycle Time(Trc) = 7
Row Refresh Cycle Time(Trfc) = 17
Row to Row Delay(Trrd) = 2
Write Recovery Time(Twr) = 2
Write to Read Delay(Twtr) = 1
Read to Write Delay(Trwt) = 2
Refresh Period (Tref) = 3120
Write CAS Latency(Twcl) = Auto
DRAM Bank Interleave = Enabled
Skew = AUTO
Skew Value = 0
DRAM Drive Strength = Level 7
DRAM Data Drive Strength = Level 2
Max Async Latency = 7ns
Read Preamble Time = 5ns
Idle Cycle Limit = 256
Dynamic Counter = Enabled
R/W Queue Bypass = 16x
Bypass Max = 7x
32 byte Granularity = Disable(8burst)
--------------------------

Super Pi
(FYI, 2x256MB will be slower than 2x512MB at every test)

1M = 29.703 s
2M = 1min 08.015
8M = 5min 25.625 s
16M = 12min 27.547 s
32M = 27min 13.765 s

Hexus Pifast = 42.31 s

2x eVGA 6800U PCI-E SLI @ stock 425/1100
(check out the cpu scores

)

3dmark2001 = 34,033
3dmark2003 = 22,986
3dmark05 = 10,527

here's some SSE3 superpi

4M = 2min 25s
8M = 5min 16s
16M = 12min 12s
32M = 26min 51s

remember 2x256MB is slower than 2x512MB heh

2x eVGA 6800U PCI-E stock 425/1100 with 72.03 driver
Aquamark3 = 96,565

Cinebench2003
CPU score = 389

Post by **Apoptosis** » Thu May 19, 2005 10:42 am

what about sandra scores?

infinitevalence · Post by **infinitevalence** » Thu May 19, 2005 10:45 am

UNFU**ING REAL !!! you hit 275fsb on BH-5 and at 2-2-2-6 timmings. You may have the highest performing memory i have ever seen.

Post by **Apoptosis** » Thu May 19, 2005 11:12 am

nah I'm running 275FSB 1:1 on BH-5, CH-5 on the DFI boards.

eva2000 · Post by **eva2000** » Thu May 19, 2005 11:21 am

yeah 2x256MB BH-5 can easily hit 275mhz - these sticks are good for 283mhz 2-2-2-6 at least

Apoptosis wrote:what about sandra scores?

Yeah san diegos have lower bandwidth than clawhammer/winnie/newcastles heh

infinitevalence · Post by **infinitevalence** » Thu May 19, 2005 1:26 pm

damn my 512mb sticks.

eva2000 · Post by **eva2000** » Sat May 21, 2005 4:16 am

well upped the multiplier for some superpi runs with 2x 512MB OCZ PC3200 Gold UTT BH-5 with 414-3 bios and orange dimm slots

@267mhz 1.5-2-2-6 7-17-3-2-1-3 at 3.64v

Bwall · Post by **Bwall** » Fri Jun 10, 2005 4:56 pm

Back again with a little something for Nate.
http://birch.net/~lindsayf/com100fps.JPG

Told you I'd get it!!! ;)

It looks fishy that it's dead on 100.0fps but that's what it does.

Here is the CPU-z screenshot of my settings. Notice that I'm running 2t and 2GB RAM. :D
http://birch.net/~lindsayf/cpuz28.JPG

241x12 1.375v +113%
2-2-2-6-2t 3.2v (mostly auto)
4x512 TwinMOS UTT

Water cooling will be here next week and I'm going to pop the top off this CPU (thanks for the pics Yves!) and see what we can do.

Currently happy with 2.6GHz and 2GB RAM.