AMD Processors
Decrease font size
Increase font size
Topic Title: slow 533Mhz K6-2 running Win XP
Topic Summary:
Created On: 12/20/2005 10:30 PM
Status: Read Only
Linear : Threading : Single : Branch
<< 1 2 Previous Last unread
Search Topic Search Topic
Topic Tools Topic Tools
View similar topics View similar topics
View topic in raw text format. Print this topic.
 03/09/2006 05:59 AM
User is offline View Users Profile Print this message

Author Icon
WarlordOne
Junior Member

Posts: 7
Joined: 01/31/2006

I've done a little more tweaking and managed a higher clock. Here's where I stand right now:

475@560 Mhz (4.5x124fsb)
Dhrystone ALU 1138 MIPS
Whetstone FPU 553 MFLOPS
Integer MMX 2531 it/s
Float 3DNow! 3011 it/s

Your max FSB is limited to 100 per Compaq's website with the highest processor support being 550MHz K6-2 or K62+. It may have additional setting not printed on the MB. But you need the actual MB model # in order to possibly find a manual. I was able to find mine in Cpu-z, Sis sandra, and it was also in the boot screen ( but I didn't know what 5SMM meant at the time). The 2x multi MAY work for you but there are no guarantees (mine worked but was unstable)
 03/21/2006 11:49 PM
User is offline View Users Profile Print this message

Author Icon
ocmaster
Junior Member

Posts: 2
Joined: 03/21/2006

quote:

Originally posted by: WarlordOne
I've done a little more tweaking and managed a higher clock. Here's where I stand right now:

475@560 Mhz (4.5x124fsb)
Dhrystone ALU 1138 MIPS
Whetstone FPU 553 MFLOPS
Integer MMX 2531 it/s
Float 3DNow! 3011 it/s

Your max FSB is limited to 100 per Compaq's website with the highest processor support being 550MHz K6-2 or K62+. It may have additional setting not printed on the MB. But you need the actual MB model # in order to possibly find a manual. I was able to find mine in Cpu-z, Sis sandra, and it was also in the boot screen ( but I didn't know what 5SMM meant at the time). The 2x multi MAY work for you but there are no guarantees (mine worked but was unstable)




quote:

Originally posted by: FieroGT42
I don't have time to read the whole thread tonight (it's after midnight) but if it wasn't mentioned here already, FSB can be much more important that raw clock speed. Check this out:
http://www.cpu-central.com/Art...p?ar...79&decor_int=23

QUOTE (WarlordOne @ Feb 4 2006, 09:56 AM)
===========================
Taz, thanks for your input as well, makes me feel like I'm kinda doing things right. I replaced the bubblegum with AS5 and confirmed the "AFX" status of my chip. So far I've run 496MHz(124*4), 500MHz(5*100), 504MHz(112*4.5), 525(95*5), 550(100*5) and oddly enough 496MHz and 504MHz seem to run the programs I use the smoothest. So, right now I'm at 504 and satisfied that this machine will burn through the web as fast as it ever has!

Thanks again!
===========================

^^^
WarlordOne, That's because you have a faster Front-Side Bus!
Anyway, I ran across this topic after overclocking the 533MHz K6-2 to 550 with only a 0.1 degree © increase in operating temp using the stock case and cooling. (I finally re-examined my old Compaq 7470's mobo and found the clocking jumper settings printed on the board, after previously giving up when told that it didn't have such jumpers )

My Base = 5.5 x 97, SiSoft Sandra yielded 1047 mips, 620 mflops
Stepping up to 5.5 x 100MHz yielded 1074 mips, 650 mflops

Overclock that 7470's FSB!

I want more on mine, but FSB settings appear to be limited to 100MHz? I will NOT sacrifice bus speed for raw CPU clocks So the 2.0 stepping is actually 6.0? I read this somewhere else and wanted to verify before I fry anything with my experiments. And can I go 6.0 x 100MHz with the stock cooling system? Is there a jumper setting on the Compaq 7470 motherboard (the "Camaro" setup) that will give me 112MHz FSB?

And finally, here's my contribution to make up for the thread hijack: For anyone else with a 7470 with the "Camaro" motherboard, jumpers 4-8 are right next to the power connector, jumpers 12-15 are where the DIMM slots (RAM) and PCI slots comem close to making a 90 degree angle, and here's what's printed on the "Camaro" mobo:

JP4 JP5 JP6 CPU/Host Ratio <--- This is the stepping multiplier
--------------------------------
On On Off 2.5
Off On Off 3.0
On Off Off 2.0
Off Off Off 1.5/3.5
On On On 4.5
Off On On 5.0
On Off On 4.0
Off Off On 5.5
JP15 JP14 JP13 JP12 CPU <--- This is the Front-Side Bus speed
-----------------------------------------
Off Off Off On 66.8
Off On On Off 95
Off On Off Off 97
Off On On On 100


 03/22/2006 12:06 AM
User is offline View Users Profile Print this message

Author Icon
ocmaster
Junior Member

Posts: 2
Joined: 03/21/2006

compaq 7470 overclocked to 600mhz tell ya how and safley 672 possible but will crash

im34cool@earthlink.net

you are able to change the multiplyer to 6 buy using the 2.0 setting for the bus speed and the core voltage must be set to 2.8 or you will have a blank screen and the bus at 100 mhz and you will be running at 600, the bios only sais 550 but after it boots and you go to control panel then system it will say 600 mhz you must have a copper heatsink or you will overheat the cpu. stable for 1yr now people say you cant run the core voltage passed 2.3 or 2.4 yah right 2.8 baby! i have also hears max hard drive only 120, ya right two 250 mb. i use 4096 on each as memory also can give you a site where you can get 533 cpu new for 13.97 if you need help call 330 264 0868
 03/25/2006 09:17 AM
User is offline View Users Profile Print this message

Author Icon
tazwegion
Senior Member

Posts: 273
Joined: 02/03/2006

Above 2.5v killed my K62-533... IMHO 2.3v ~ 2.5v is absolute maximum for a 2.2v core processor

-------------------------
 10/01/2006 01:35 PM
User is offline View Users Profile Print this message

Author Icon
Suzueya
Member

Posts: 135
Joined: 09/25/2005

Personally I wouldn't go over 2.4v, though that doesn't mean it can't be done. Another consideration when picking the settings you want to run at is the fact that the 5.5 multiplier gives noticably better results in terms of memory speed operation than any of the others do. i.e 5.5 x 103 works better than 6 x 103 for best memory speed. Why ? I don't know; but it is a fact!! Come down to it, 5.5 x 100 works better than 6 x 103 or even 6 x 105 for memory write speed.

Superpuppy 3 DFI K6BV3+/66 (1 Meg Cache) w/ K6-3 450ACZ @2.1v, & 2x 256 Meg Hynix PC133 CL3 SDRAM (CL2 @ 100 - 103 MHz)

Edit : 13/08/05 5:22 PM - New Test Results for Superpuppy 3
K6-3+ 450ACZ @ 2.0 v. @ 5.5 x 100 MHz w/ latest WP Credit tweaks.
Everest :
Memory Read = 303
Memory Write = 158
Memory Latency = 211.8
Sandra :
Int MMX = 210
Float FPU = 208

Edit : 24/10/05 7:03 AM - New Test results for Superpuppy 3
K6-3+ 450ACZ @ 2.1v. @ 5.5 X 103.2 MHz w/ CPU Cool w/ same WP Credit tweaks as above.
Everest :
Memory Read = 312
Memory Write = 164
Memory Latency = 206.4
Sandra :
Int MMX = 213; 214; 216; 216; 216.
Float FPU = 212; 213; 214; 215; 214.

EDIT : 24/10/05 8:02 AM New test results on Superpuppy 3
K6-3+ 450ACZ @ 2.1 v. @ 5.5 x 103.2 MHz w/ CPU Cool, raised to 6 x 103.2 MHz by K6Speed w/ same WP Credit tweaks as above.
Everest :
Memory Read = 312
Memory Write = 150
Memory Latency = 202.1
Sandra :
Int MMX = 214; 215; 215.
Float FPU = 213; 214; 213.

Would note in passing that even those boards which have no provision for FSBs higher than 100 can in some cases still have their FSB overclocked beyond 100 MHz by using software to do it. My DFI K6BV3+/66 only has settings for a max FSB of 100 MHz; but I have used "CPU Cool" to clock the FSB up to 112 MHz, though that caused problems because my RAM would not support the timings I use at that speed. 103 MHZ FSB works fine though.

Superpuppy 2
K6-3+ 450 ACZ (6x100) - (For now) w/ 48mm CoolerMaster (Max Possible)
ASUS P5A-B Rev 104 (512k) w/ 2x 28mm Chipset Fans
3x256 Meg (mismatched) PC133 SDRAM (Reduced to 2x for these tests)
2x 40G WD (7200) HDDs Non Ducted w/ 2x486 Exhaust fans right behind
1x20G Maxtor (5400) in hot swap tray (Contains Backups)
1x Hot Swap Tray, gutted and modernized w/ ATA 133 internals w/ fan included
1xLG 52/24/52/16 CDR/RW/DVD ROM
32 Meg ATI aiw Radeon DDR AGP
SB Live 5.1 Sound
Olicom 2527 Lan
1x80mm Front Chassis Fan & 2x486 Front Chassis Fans

Superpuppy 2 : K6-3+ 450 ACZ @ 5.5 x 100 MHz @ 2.1 v.
Everest :
Memory Read = 317 MBs
Memory Write = 153 MBs
Memory Latency = 201.8 ns
Sandra :
Int MMX = 198
Float FPU = 195

Superpuppy 2 : K6-3 450 ACZ @ 6 x 100 Mhz (600 Mhz) @ 2.1v
Everest :
Memory Read = 318 MBs
Memory Write = 140 MBs
Memory Latency = 204.2 ns
Sandra :
Int MMX = 196
Float FPU = 193

Superpuppy 2 : K6-3+ 450 ACZ @ 5.5 x 105 Mhz (577 Mhz) @ 2.1v
Everest :
Memory Read = 340
Memory Write = 165
Memory latency = 194.7
Sandra : - (Because Sandra results rise with successive testing, I ran the tests repeatedly until the results started to taper off)
Int MMX : 196, 197, 212, 215, 222, 224, 223.
Float FPU : 195, 201, 212, 217, 221, 220, 220.

Superpuppy 2 : K6-3+ 450 ACZ @ 2.1v. @ 5.5x105 (577 MHz) raised to 6 x 105 (630 MHz) by K6Speed.
Everest :
Memory Read = 336 MB/s
Memory Write = 141 MB/s
Memory Latency = 197.5 ns
Sandra :
Int MMX : 195, 195, ***, 212, 217, 219, 218.
Float FPU: 195, 200, ***, 213, 216, 215, 215.
*** = Test Failed over half way through. Popup window : "This Program Has Performed a Diseased Avian Operation And Will Now Close". Sandra closed but other than that things were ok.

You will notice that the "Memory Write" results were higher @ 5.5 x 100 Mhz (550 Mhz) than they were at 6 x 105 Mhz (630 Mhz). What all this confirms is that there is a definite performance advantage tied to using the 5.5 Multiplier unadjusted. Obviously something has been optimized to work best w/ the 5.5 multiplier.

Having installed three 256 Meg sticks of Hynix HYM71V32635HCT8P-K, (PC133 CL2), in Superpuppy 2, I decided to see if I could get 112 MHz FSB stable.

Superpuppy 2 : ASUS P5A-B Rev 104 w/ K6-3+ 450 ACZ @ 5.5 x 112 MHz @ 2.3v. w/ mobo cache enabled. (First attempt; - has gone well, will try to reduce to 2.2v. later).
Everest :
Memory Read = 361 MB/s
Memory Write = 177 MB/s
Memory Latency = 182.8 ns
Sandra :
Performance Rating (Estimated) = 740
Arithmetic = 1521 D ALU MIPS ; 731 W FPU MFLOPS.
Multimedia = 3544 INT MMX ; 4217 FLOAT 3D NOW!
Memory Bandwidth : (Run repeatedly on account of rising results)
INT MMX : 210 ; 213 ; 229 ; 231 ; 238 ; 241 ; 240.
FLOAT FPU : 210 ; 216 ; 228 ; 234 ; 237 ; 237 ; 237.
Cache & Memory Bandwidth : 1164 - 2k ; 1179 - 4k ; 1183 - 8k ; 1164 - 16k ; 1120 - 32k ; 1036 - 64k ; 915 - 128k ; 630 - 256k ; 268 - 512k ; 236 - 1Meg ; 236 - 4Meg ; 235 - 16Meg ; 237 - 64 Meg ; 236 - 256Meg.

Rerunning Everest AFTER Sandra gave the following results :
Memory Read = 365 MB/s
Memory Write = 201 MB/s
Memory Latency = 184.3 ns

All tests run under WIN98-SE ; w/ Power Tweak 2's tweaks ; w/ : CAS Latency = 2T ; RAS to CAS Delay = 2T ; RAS Precharge = 2T ; RAS Active Time = 6T ; Row Cycle Time = 9T. - & my usual mess of 10 taskbar items running.

Settings, = Customized Powertweak 2.

Powertweak 2 Processor Settings :
Write Allocation = Enabled.
Write Combining For VGA Framebuffer = Disabled.
Write Combining For VGA Banked Memory = Disabled.
Data Prefetch = Enabled.
Weak Write Ordering = Disabled.

Powertweak 2 ALi Chipset Settings :
Use M1541 internal TAG RAM = Disabled.
Use M1541 internal MESI = Enabled.
Cyrix Burst Mode & K6 Write Allocation = Enabled.
Cyrix Linear Burst Order = Disabled.
L2 TAG Output Delay = Enabled.
CPU Single Read Cycle L2 cache Allocation = Enabled.
Dynamic Write Back = Enabled.
Pipeline Function Control = Disabled !!!
Write Combining For VGA Fixed Framebuffer = Disabled.
Write Combining For Linear Framebuffer = Disabled.
SDRAM CAS Latency = 2 Clocks.
SDRAM RASJ Precharge Time = 2 Clocks.
SDRAM RASJ Cycle Time = 9 Clocks.
A.G.P. Side Band Addressing = Enabled.
A.G.P. Transfer Rate = 2x.

Anyway, thought I'd give this thread a bump so that anyone who has checked the "SS7 evolution" link Taz stuck in can see the latest mods to the "Dog".
 10/18/2006 11:07 PM
User is offline View Users Profile Print this message

Author Icon
deCat
Junior Member

Posts: 7
Joined: 10/18/2006

Been messing with K6/2's since 1999 (& K6/3's later) & love playing with them.
Have even done mpeg2 video encoding on them (slow, but the output is fine).
Got 2 such systems (1 w/ K6/2@500 & one with a K6/3-350 OC'd to 400MHz) over on the wall running right now (LAN fileservers & occasionally video edits when I got overload of edits to be done).
Both carry V3-3000 video, one AGP (the k6/3) & the other PCI.
Both with W98gold as OS


Saw this thread while digging on other things & decided to add a few comments.

If you read AMD tech docs on these cpus carefully, you will find that there are 2.4V & 2.2V cores up to 533MHz (AFAIK all 550's were 2.3V rated).

The 2.2's & 2.3's have a 'book' absolute top voltage of 2.5V, & the 2.4V jobs top at 2.7V.
*I* would highly recommend that these not be exceeded unless you have much better than stock cooling for the chip!


ALSO, my current K6/2 sys uses an IBM MB (from a sys I bought retail in 1999 as a K6/2-475) & has absolutely NO room to get a better HSF unit on it, barely has clearance for the stock unit.
Ran that sys @ 550MHz for a couple of years (have 550, 533, 475, & 450 chips that will all OC to 550 fine).
But was plagued by random BSOD's (could go weeks at times, other times would get them on boot or reboot).
Didn't matter which chip of the 4 was in either.
Can't remember the exact error reported on the Blue screens; vmm (something) .vxd; but finally did a google search on that error & ran across 2 reports of 'known' heat issues with K6/2's at 550MHz.
Had the 550 itself in at the time, & pulled it back to 500MHz have never seen that screen since; have had 98 up "active" for as long 4 months w/o a reboot (used it as both fileserver & a private FTP server at the time, along with fairly heavy use of an FTP client).
BTW, for those who think W98 has to be reloaded/reinstalled every so often:
I use multiple regcleaners, & do manual regedits to remove left over entries when uninstalling s/w.
& do a lot of 98gold tuning on my 98 installs.
That OS has been moved across at least 6 different HDD's, but is STILL the orig IBM OEM one (minus the IBM preload garbage of course)!
It's been restored via ghost image a couple times in cases of HDD failures, but it's still the orig OEM OS none the less.
& that OS still runs fine w/o bogging down.
- - - - - - - - - - -

Also: have done a lot of benchmarking & performance checking over the years; even "real world" tests like doing the same video edit under different settings.
& I can confirm that a K6/2 will always give best performance at the highest reasonable FSB speed.
Logical really; a /2 has it's L2 cache on the Motherboard & accesses that via the FSB!
A k6/3 is slightly less bus limited, but generally will also do best at highest FSB because even tho it has 'onchip' L2 it can make good use of the motherboard's cache as L3.

Ya really want to see a /2 or 3/ 'talk', get ahold of one of the rare MB's that has 2meg onboard cache.
I've seen one & ran tests on it, but couldn't talk the guy who had it out of it!!!
Been so long I can't remember the brand of the board; just that it came with 1meg onboard standard & had a socket another 1 meg chip could be added with. Cost the guy almost as much as his 450 CPU to get that extra L2 chip, but was worth it in performance (for one thing, it allowed caching of all the RAM instead of just 128meg of it like most standard 512K L2 setups)
= = = = = =


On another note:
My K6/3 is on an ASUS P5A MB (nice board, 3 ram slots, AGP slot, & etc); can't get it past 400MHz no matter what voltage used, have been all the way to 2.9 before I gave up. Got a nice coolermaster K7 HSF unit on it (barely fits in) so that one is well cooled.

Got my hands on a new K6/2+ 450MHz a couple years ago & tried it in that board (have last BIOS released & etc).
Can get the chip to 600MHz easily, boots fine & shows in BIOS & windows as being @ 600MHz, even in CPU-z; problem is that I have a Later revision of the board with a known h/w issue that causes the chip to work full time just running the board with little cycles left over for 'user' performance! :-(

One of these days when I get time (currently have a 1333 t-bird that killed it's boot HDD & a second A64 in final setup & tuning) I got ahold of a Compaq MB that is supposed to run a 2+ OK to try that chip with.
note: I have had access to my mother's PII-400 IBM (bought just before I got my K6/2) all along (sole access now because she's in a nursing home with alzheimers) & the little K6/3 @ 400 cuts that one NO slack performance wise.
So I'm anxious to know how well that k6/2+ will run when I can open it up properly @ 600MHz!


K6/2's & /3's are the chips that made an AMD man outta this old f*rt!
Got 5 AMD systems on my LAN right now in fact, 4 desktps & a NoteBook. ;-}
Love ta see that others are still messing with the old chips too!
 10/28/2006 12:16 AM
User is offline View Users Profile Print this message

Author Icon
Suzueya
Member

Posts: 135
Joined: 09/25/2005

@ de Cat : I have 3 of those 2Meg cache boards. One is being used in Superpuppy 3 in its current rebuild, the other two are spares. The one being used : DFI K6BV3+/66 Rev B2 w/ 2Meg cache. The two spares : DFI K6BV3+/66 Rev B1 w/ 2 Meg cache. When all the mods are done and the rebuild is finished , I'll post some new results. To see the mods, click on the "SS7 Evolution" link earlier in this thread.
 10/28/2006 05:52 PM
User is offline View Users Profile Print this message

Author Icon
NyteOwl
Member

Posts: 79
Joined: 08/25/2006

I have XP running fine on a 550 MHz Celeron/128M in my laptop and on a K6-III+/450 @ 550 w/512M on an OEM Tyan version of the Compaq (Camino) motherboard. The trident video is not very good, maxing out at useble resolution of 1024x768. I found an ATI 7500 Radeon 64MB PCI (no AGP slot) that does a nice job.

The sister to this system is a K6-2/500 w/512M and an ATI 7000 Radeon 32MB PCI running Slackware Linux flawlessly.

I might not want to do a 3D render on these (or compile KDE from scratch - 4 days) but they are quite functional and acceptably fast for most uses.



-------------------------
Obsolescence is just a lack of imagination. 36 Bits Forever! #include <disclaimer.h><br>Milner Manor | The Yak Shack | Caisteal Comraich
Statistics
112018 users are registered to the AMD Processors forum.
There are currently 0 users logged in.

FuseTalk Hosting Executive Plan v3.2 - © 1999-2014 FuseTalk Inc. All rights reserved.



Contact AMD Terms and Conditions ©2007 Advanced Micro Devices, Inc. Privacy Trademark information