Hey, that realtime benchmark is really cool! Now I can see how slow my mouse interrupt is, the score drops from a steady 447 to about 300 when I move it 😳

My first thought is "serial mouse" -- am I right?

My second thought is "I hope he wasn't submitting timings with the mouse driver loaded, you're supposed to run the program in a clean boot!" 😉

No, it's a PS/2 mouse, and yes, I had all standard TSRs loaded. On a clean boot the score increases to 448-450.. And moving the mouse won't change it. As expected.

But my point was, that the mouse interrupt handler on my system is very slow.. I noticed it before but now I can see how much the system actually slows down when I move it.

I stand corrected. Was that cutemouse as your handler? If not, what mouse driver has such poor code?

Seeing my M919 AM5x86 160MHz monster equaled closest to a Latitude DX33 makes me feel inadequate. A score of 125 🙁

Bad video score maybe?
My M919 went up quite a bit switching to an S3 VLB card from a PCI Mach64.

😮

Well, it got worse witn another video card 😜
one so bad, I can't even match the speed of the PCem emulator to these two results

Something must really be wrong then 😳.
Here's mine:

Yeah, now you're making my M919 look very strange even if it does have the same am5x86 (also incorrectly reported as AMD Enhanced 486), lack of L2 cache and an s3 card. 25 3dgames score!?!

This is another thanks to Trixter here for helping me realize how crap my M919 is to other M919s 😀

So what's a good start toward this quality result? Achieve 150MHz instead of 120/160?

Hmm, I don't know about that.
I lost some points going back to stock from 50MHz, but not that many.

Maybe one of us is getting incorrect results?

Dunno. Got to 150mhz, put the S3 back in, seeing 131 topscore, 3dgames is 89.

EDIT: Strange. Did "Load Optimal Defaults" and i'm now up to 230 topscore and 3dgames is 35. I wonder what was slowing me down 😖

Ok I have some new scores:

Another 386DX-40 and a Compaq Presario CDS520 with a problem.

1;AT clone 386 home built
2[UID62512960]
3MemoryTest=161
4OpcodeTest=59
5VidramTest=160
6MemEATest=50
73DGameTest=56
8Score=106
9CPU=AMD Am386DX
10CPUspeed=40 MHz
11BIOSinfo=R(C)1985-1993,American Megatrends Inc.,All Rights Reserved,6145F Northbelt Parkway,GA-30071,USA.(404)-263-8181. (08/08/93, rev. 0)
12BIOSdate=19930808
13BIOSCRC16=6251
14VideoSystem=VGA
15VideoAdapter=VGA, Cirrus CL-GD5428, VESA, 256kb Video Memory (BIOS)
16Machine=AT clone 386 home built
17Description=386DX40/ULSI83C87/16MB/Ali1429/V7VEGA1MB
18Submitter=Michael Huth
19
20;Compaq Presario CDS520
21[UID904FE1032]
22MemoryTest=90
23OpcodeTest=48
24VidramTest=134
25MemEATest=72
263DGameTest=36
27Score=177
28CPU=Am486SX2-66
29CPUspeed=66 MHz
30BIOSinfo=\CL GE S CAB2/NPBCWKDJWDJCDSJCGCMRGTRZAZA(C)Copyright COMPAQ Computer Corporation 1982,94* (10/11/94, rev. 0)
31BIOSdate=19941011
32BIOSCRC16=904F
33VideoSystem=VGA
34VideoAdapter=Cirrus Logic CL-GD5424 512kB
35Machine=Compaq Presario CDS520
36Description=Am486SX2-66/52MB/CL-GD5424/ELNK3/ESS688/GUS ACE
37Submitter=Michael Huth

TOPBENCH freezes on CPU Speed detection for the Am486SX2-66. Realtime Benchmark works ok though. I had to run it with -s and entered the data manually.
Maybe noteworthy, MemEA on the 486 has a slower value than on the 386.

Thanks for the results, I've added them and will update the website soon (I have a new release to publish that adds some minor functionality). Your 386-40 is the fastest recorded, btw.

As for the CPU detection, it's dodgy and fails on a handful of systems, hence the ability to disable the detection and enter the data yourself. I guess you got "lucky" with the compaq, sorry.

Interesting regarding the EA going down on a 486 -- I checked a few sources included http://www.phatcode.net/res/224/files/html/ch12/12-02.html and have concluded that the metric code "violates" rules that optimize such calc on 486s. (Of course, there's nothing I can do about it since the benchmark can't change or all of the database results will become invalidated.) Interesting, nonetheless.