Here are my DX50 results (submitted to Phils VGA database):
3Dbench - 31.5
PCPbench - 8.4
DOOM - 18.42
Quake - 5.1

Hardware:
Intel 486DX-50
AUVA CAM50-UO (UMC UM8498F chipset) (ISA only)
8MB RAM
ATI Mach32 2MB VRAM ISA

I'd say yours look very much on the spot.

With the DX-50 the fsb and the internal clock should both be 50mhz. A DX2-50 would be 25/50. I wouldn't build a DX-50 system, myself. Early boards that run at 50mhz are unstable and later boards that support 50mhz are better off with a faster 486 or 5x86 in them.

sliderider wrote:

With the DX-50 the fsb and the internal clock should both be 50mhz. A DX2-50 would be 25/50. I wouldn't build a DX-50 system, myself. Early boards that run at 50mhz are unstable and later boards that support 50mhz are better off with a faster 486 or 5x86 in them.

I agree but seeing as I already have several 486 systems built (33mhz, 66mhz, amd 5x86 133mhz and Crrix 120mhz 5x86) I thought a 50mhz machine would be a fun little project seeing as I already had a spare 66mhz PC just sitting around.

vetz wrote:

Here are my DX50 results (submitted to Phils VGA database): 3Dbench - 31.5 PCPbench - 8.4 DOOM - 18.42 Quake - 5.1 […]
Show full quote

Here are my DX50 results (submitted to Phils VGA database):
3Dbench - 31.5
PCPbench - 8.4
DOOM - 18.42
Quake - 5.1

Hardware:
Intel 486DX-50
AUVA CAM50-UO (UMC UM8498F chipset) (ISA only)
8MB RAM
ATI Mach32 2MB VRAM ISA

I'd say yours look very much on the spot.

Seems on par with my DX50. How much do you get in speedsys?

Programs that can fit inside the CPU's internal cache will run at the same speed on the DX-50 and the DX/2-50. Software that is I/O intensive should get a pretty nice speed boost with the DX-50 (since in theory your memory throughput should be a lot higher). However, this is not always the case in practice. A lot of motherboards can't really handle 50MHz the way it was meant to be run and cheat by adding a bunch of extra wait states. On many of my SiS boards for example, tightly tuned 33MHz memory outperforms 50MHz memory with a bunch of wait states.

Still, VLB at 50MHz should be pretty bloody fast, especially if you can do it with 0WS.

Anonymous Coward wrote:

Programs that can fit inside the CPU's internal cache will run at the same speed on the DX-50 and the DX/2-50. Software that is I/O intensive should get a pretty nice speed boost with the DX-50 (since in theory your memory throughput should be a lot higher). However, this is not always the case in practice. A lot of motherboards can't really handle 50MHz the way it was meant to be run and cheat by adding a bunch of extra wait states. On many of my SiS boards for example, tightly tuned 33MHz memory outperforms 50MHz memory with a bunch of wait states.

Still, VLB at 50MHz should be pretty bloody fast, especially if you can do it with 0WS.

hmmm, there is a jumper on the MB to set VESA wait states between 1 and 0 but I get the exact same bench numbers regardless of the setting. everything in BIOS is set to the turbo or fastest settings and so far its been completely stable

Vetz, does your l2 cache work ok at 50MHz fsb? Can anyone give an example of a game or program where the dx50 beats out the dx2-66? In every bench I run the dx2-66 smokes the dx50 in this setup.

soviet conscript wrote:

Vetz, does your l2 cache work ok at 50MHz fsb? Can anyone give an example of a game or program where the dx50 beats out the dx2-66? In every bench I run the dx2-66 smokes the dx50 in this setup.

Yes, no issues encountered. I can't think of any at the moment. I've bought a DX2-66 Overdrive which will fit into this board so I can do a comparison.

Btw, if your DX-50 is running VLB I'd recon you would smoke my ISA VRAM based system.

Here's a video I did comparing a few cpus in the exact same machine. I don't think the l2 cache is running though with the dx50, cachechk crashes when I try and run it. Otherwise it runs fine. https://youtu.be/PHggOR-6Zig

soviet conscript wrote:

Here's a video I did comparing a few cpus in the exact same machine. I don't think the l2 cache is running though with the dx50, cachechk crashes when I try and run it. Otherwise it runs fine. https://youtu.be/PHggOR-6Zig

You do have 16 more mhz which may provide better results in 3DBench than increasing the FSB. Could be different in other benchmarks.

I currently dont have my DX50 system operational, so can't test this at the moment.

A few reference values :

(MB1433/40/50 UIV-A rev 6, DC680T VLB cache controler, 32MB ram, ET4000/W32p VLB 😀 )