First post, by melbar
- Rank
- Oldbie
I've planned to build a Retro PC over a half year ago. I wanted to re-activate my old dual-celeron system (2x466MHz) on the famous Abit BP6 board. I used this system around 1999 to 2002. The former Voodoo3-2000 and the Geforce256 DDR, which were the GPU cards for this system, doesn't exists anymore cause they were scrapped long ago...
But anyway, i was not able to boot this old hardware, the old air coolers were also really loud, and finally i've decided to sell it on ebay. As a good replacement, i've bought really cheap a Pentium III 866 system + Geforce3-Ti200.
During last chrismas, i've found the really interesting video's from Phil'scomputerlab on youtube about building Retro PC's. And especially the video about the '4in1 time machine' was really inspiring me. Then i've decided to buy in addition a super-socket 7 (SS7) board for Pentium's and AMD K6's processors.
The purpose of this SS7 board for me was, to run the PC also with the lowest speed that a 386 or 486 system is not really necessary anymore, due to the reasons:
1.) nowadays depending on board + CPU this old hardware is really expensive and
2.) the problem which can occur with AT-formfactor, AT-power supply and often with the RTC-clock.
I know that mainly the lowest FSB of a SS7 board is 66MHz. That means a Pentium 75MHz or 90MHz can't run. And as it comes, the board that i've bought comes with a K6-2 500 CPU.
So, the first idea was: Ok, let's clock this CPU with the minimum multiplier 1.5x and have fun with 100MHz. Written in manual and also as a table onboard the mainboard, the smallest multipliers written as '1.5x/3.5x'
It turns out that the K6-2 takes the higher one...And also the 2x muliplier does not work cause the 'modern' K6-2 CXT interprets the second multiplier as 6x.
Then, i was really thinking that with the resultant clock (2.5x66MHz = 166MHz) the system becomes too fast. So, i've bought (also really cheap 😀) a Pentium 150MHz for a final clock of 100/133 MHz.
This weekend i've completed all the tests which i need for a good impression of CPU's from PI+PII+PIII and AMD K6 /-2 /-3 era at their lowest possible speed. My tested hardware with Windows 95 is:
- AMD K6-2 500
- Shuttle HOT597 V14 (with VIA's 82C598MVP chipset)
- 32MB PC66 or 64MB PC100 SDRAM
- PCI ATI Rage XL 8Mb
(During the test, i was only able to set the min. FSB to 68MHz. Even with the setting 66MHz, the cpu runs with 68MHz FSB. Therefore the absolute minimum clock is 171MHz. That means, all clocks of my K6-2 looks not usual due to the 3% overclocked FSB.)
- Pentium III 866
- QDI SynactiX 5EP-A Rev2.0 (with Intel's 815EP chipset)
- 256MB PC133 SDRAM
- PCI ATI Rage XL 8Mb
The last days before the tests, i've tried to predict (calculate) the 3DBench & PCPBench values of a K6-2@166MHz and a Pentium @100MHz / 133MHz, with the values from Phil's K6-III+@133MHz and the corresponding benchmark values from cpu-world. Well, when the architecture of a cpu is similar and the benchmark scale's linearly, then those interpolated values can match the test values without big gaps. But as my testet CPU's have different architecture and also with the scenario's of disable the caches, all this becomes non-linear!
The benchmark, i've only used are the following:
- 3DBench (from Phil's package)
- PCPBench (from Phil's package)
and only a game bench (for live feeling):
- running Wing Commander (like in Phil's video's with testing 386-DX40 / -DX33 / -DX25 Speed).
I have attached all the data (tables + charts) from my tests as png files.
Note:
I know that the values at full speed can be higher for the K6-2 and PIII, even for the coppermine. But in these test's i didn't use modifications like SETK6 or MTRRLFBE or FASTVID.
Conclusion:
As i did not expect: The K6-2 @171MHz is even more slowly during the disable of L1 and L1+L2 caches (for the 486 and 386 mode), compared to the Pentium @137MHz or @103MHz. That means the Pentium is really faster regarding the speed per clock cycle compared to the K6-2. This has also something to do with the different build of these CPU's: the Pentium has a CISC core and the AMD has a RISC core with x86 to RISC86 decoder. The AMD structure is more influenced by disable the caches.
For the Pentium III @431MHz (66MHz FSB): I am really surprised about the behavior of the coppermine when disable the cache and comparing the two benchmarks. First, it doesn't matters if you disable only L1 cache or both L1+L2 cache. The results are the same. With the 3DBenchmark i've got nearly twice the value of the Pentium@137MHz with L1 disable. But the PCPBenchmark, the Pentium III undermatches all the settings of Pentium & K6-2. Really interesting, i don't know why this can happen...
#1 K6-2/500, #2 Athlon1200, #3 Celeron1000A, #4 A64-3700, #5 P4HT-3200, #6 P4-2800, #7 Am486DX2-66