? for Peter Missel if you would please ?

[scotter]

as you can see from my other post's I've clocked the you know what out of my chip and made some very none stock cooling additions to it http://www.sysopt.com/forum/smile.gif
any way this is the problem windows and all of my programs run fine I do not notice any slow down at all but when I benchmark this chip the cpu and fpu scores are way low
at 550mhz in wintune I'm getting

Summary
(1) 551 MHz
218 MIPS (Integer operations)
360 MFLOPS (Floating point operations)
20 (Integer application simulation)
34 (Floating point application simulation)
23 (MMX application simulation)

CPU Details

CPU load: 0
low MIPS: -1
CPUID: 0x058C 0x8021BF
MMX Present: True
3DNow Present: True
Streaming SIMD Extensions Present: False
Processor Serial Number Present & Enabled: False
dhrystone time (s): 9.1
whetstone time (s): 0.028
Integer time (s): 19
Floating point time (s): 10
MMX time (s): 20

and in sandra

cpu 285
fpu 230
mmx 1412
3dnow 1062

any Idea of what might be going on here ?
I'm at a loss to figure this one out

Also do you know what the (AMD-DRX-BUG) is ?

Thank you for any help http://www.sysopt.com/forum/smile.gif
Scotter

[Peter M]

So does the board run 100 MHz, or something above that? Above 100, BIOS might put the brakes on some components to get it stable, decreasing performance. (Note that I'm not at all into benchmarking, and I have no idea about what a "good" or "low" Sandra benchmark would be. Benchmarking a processor from inside a multitasking environment can't be accurate anyway ... and since Sandra gave me so many incorrect reports on system configuration on the few occasions I used it, I don't mind its results anymore.)

My current favorite for memory bus throughput is CACHEMEM.EXE (run from DOS), this gives pretty accurate results. This is about the only figure that makes a difference in overall CPU performance, since the CPU's internal number crunching speed, if it isn't broken, is always the same at a given core frequency, regardless of the mainboard.

And which tool detected that "AMD DRX BUG"?

Regards, Peter

[scotter]

yes it's running at 100mhz front side bus it does it at 100mhx fsb 105 mhz fsb and 110fsb I'm using a Asus P5A mobo
the program that said this thing about
AMD drx bug is called Diag.exe I downloaded it from tweekfiles.com it runs in dos
and I'll run cachemem on this thing latter today and post the results for you
thanks Peter for your input http://www.sysopt.com/forum/smile.gif

Ok where can I find cachemem at I looked all over the net and cant find it ? http://www.sysopt.com/forum/smile.gif


[This message has been edited by scotter (edited 03-03-2000).]

[scotter]

here is some more info for you if this helps
I swaped out a K6-2-300 I have a another computer and the marks show what they should with it
the other thing is I run rain when I first started using it I would see about a 10/c to 15/c drop in temps now it only drops them around 2/c to 3/c I've tryed powertweek and the software cooler in it is the same almost no drop in temps with it running.

any ideas ?
thanks, Scotter

[Peter M]

Those low figures, plus the increased overclockability of that chip, suggest to me that either the Instruction or Data L1 cache is fried.

AMD processors do a cache self test on startup, and won't use it if it doesn't work. Try ctcm.exe from ftp.heise.de, this program measures cache sizes by looking at access latencies (instead of just taking the CPU type and displaying the cache size this type of CPU _should_ have like most other tools do).

Regards, Peter

[scotter]

thanks Peter I got a copy from tomshardware but I cant get it to run on my computer it gives me a run time error 200 at 0997:0091
it runs on my othere computer but not this one ?
if I go into bios and disable the internal cache then my system slows way down and the scores drop even more to extremly low cpu 25 fpu 23 from 276, 244.
if i go in and disable external cache there is almost no diff at all that I can see cpu went from 276 to 255 fpu stayed the same
in sandra

any way thanks so much for your help http://www.sysopt.com/forum/smile.gif
Scotter.

[This message has been edited by scotter (edited 03-04-2000).]

[Peter M]

Get a newer ctcm.exe from ftp.heise.de (that's where it originates from). Current release is 1.6R.

What you wrote looks like your CPU has only Data or Instruction cache fried. They cannot be individually disabled in software, that's why you see a further big decrease in benchmark.

Regards, Peter

[scotter]

Hi Peter and once again thanks
when I go there I only see 2 zip files to down load
ls_lr.z and ls_lr.gz and I dont understand german
so what one to get if those are the ones I need ?

[Peter M]

Use that link then:
http://www.heise.de/ct/ftp/ctsi.shtml

Time to polish up your German (or go through a translation machine :-)

Regards, Peter

[scotter]

Peter your the man for sure got it and ran it thanks for the link here is what it said

Processor timing : am6k86

Processor cpuid : Autentic AMD typ=00 fam-05 mod=08 rev=0c feat=008021BF

AMD K6 config : write allocation to 128 mbytes

write combining LFB : disabled, BF-pins = 001 => ratio = 5.0

Acutal clock rate : 561.169 MHz

Primary cache (L1) Not present or not active
or >= 8 way Associative

secoundary cache (L2) not detected or greater than 2048kb

Code cache (L1) 32Kbytes, 2 way associative

Main memory 128mbytes, no memory holes found


Transfer in 4gbyte real mode no paging , via cpu integer unit

best rate for 8k mouse cache/page hit :91.1us=> 89.9 mb/s

meadium rate for 8k movd miss + hit : 90.1us => 90.9 mb/s

" " " " ( L2 clean ) 90.6us => 90.4 mb/s

" " " " ( L2 dirty ) 121.9us => 67.2 mb/s

worst rate for " " misses 112.2us => 67.0 mb/s

via FPU 8k fmoui misses 72.2us => 113.4

via MMX 8k mmoui misses 80.3us => 102.0 mb/s

blocktranfer 4m mousd 77.5ms => 54.1 mb/s

averaged under dos ( 640k ) 99.9us => 82.0
mb/s
Averaged under windows ( 4mb ) 105.0us => 78.0 mb/s

now I just need to know what it all means http://www.sysopt.com/forum/smile.gif
again thanks for all the info and input on this If there is ever any thing just ask http://www.sysopt.com/forum/smile.gif
scott http://www.sysopt.com/forum/smile.gif

[Gutter Ball]

Hi guys,

I'm just gonna jump in here and say that there is NOTHING wrong with our CPU's. It's our mobo's that's the problem. I took my 450 and put it in my godkid's 500 system and the Sandra scores were almost identical. So something is amiss with our ASUA P5A's :/

[Peter M]

... this

Primary cache (L1) Not present or not active
or >= 8 way Associative

secoundary cache (L2) not detected or greater than 2048kb

means that the primary 32 KB of D-cache are not operational, while the I-cache ("code cache" in ctcm) is. In other words, it has failed the K6 internal self test. Try clocking it back down to where it belongs, and retry to see if it's fried or just overoverclocked :-)

Regards, Peter

[Peter M]

epilogue ...

CTCM 1.6r misdetects the K6-2 or K6-III L1 D-cache. Older releases get it right ... so much for accurate benchmarks :-) (You can still override the detection from the command line.)

Regards, Peter

[scotter]

it's fried but heres the catch I swaped it with my K6-2 300 board to board and benched both chips at 333MHZ at same speeds board to board and in the EFA viking 2 board it benchs the same as the 300 I'm running at 333 4.5 X 75 @ 2.2 volts both chips bench the same.
In the Asus P5A at the same settings this one shows **** the 300 screams in 3d now its getting 1215 while this chip is showing 464 it does not matter what speed or bus or mulitplier I use this K6-2-450 sucks in this board http://www.sysopt.com/forum/frown.gif

any way I'll be getting a new chip tomarrow and we will see what happens http://www.sysopt.com/forum/smile.gif
I'll let ya know then http://www.sysopt.com/forum/smile.gif
and one more time thanks for all the input your a good man Peter Missel