We have actually touched on one of my long standing projects... still WIP as I only have 6 sticks fabricated and finished, the issue currently being free time to make more!
The 40ns, 1MB, 30 pin simm... with parity: -
A thread of the struggle... and using a different SMD method is in the pipeline.
I'm super interested in 40ns SIMMs for my 50MHz systems.
I have not done any benchmarks with them as yet, only function tested. The theory is they will boot systems with higher FSB's without need for any wait states... where conventional 60ns/70ns RAM won't even boot the system! These will also require less 25ns/20ns cycles to perform read/write operations.
Anonymous Coward wrote:
It would also be cool if somehow SRAM SIMMs could be fabricated.
I'm not sure if it is possible... would come down to control signal differences with DRAM. If there are no differences then the only barrier would be density... i.e. the real estate used per 1Mb. That problem can be solved but it would be a real PITA!
For example, and in theory... to make 1 stick of 256Kb, you would need 9 IC's = 8 IC's x 32Kb = 256Kb SIMM + parity. Most motherboards only have up to 8 SIMM slots so your total RAM would not be higher than 2Mb... From here, you would need to use these (I own a few and plan to make clones): -
They do look very good, but that might be due to a glitch. The RAM I used for the run with two 4MB modules installed were TI 60ns 4MB 3-chip modules. However, they aren't exactly 100% compatible with the board. Even though it saw 8MB, the BIOS failed the memory check at 5MB. Yes, I tried different combinations of those modules since I have 4 of them and all fail at the exact same spot. I have two 4MB 9-chip modules on the way and will retest if they work.
I like your BIOS options for cache and RAM timings. I haven't seen a "Cacheable RAM address range" option in these BIOSes before. Does it go higher than 64 MB?
Slow Refresh should be enabled for optimal performance, assuming your RAM supports this. Most do.
You mentioned that using a CPU cooler improved performance. This seems very odd to me. Could you benchmark DOOM with the cooler and without the cooler? What are the two scores?
Plan your life wisely, you'll be dead before you know it.
For DOOM timedemo 3, I get 14.1 fps. For your system with the same graphics card, you get DLC-40 = 11.3 fps, and DRx2 = 12.6 fps. With VLB graphics, you got DLC-40 = 12.85 fps, or DRx2 = 16.4 fps. Why is the performance increase for the DRx2, when using VLB, much greater than when the DLC uses the VLB graphics?
The higher the CPU, the more severe the performance bottleneck due to the ISA graphics card bottleneck.
feipoa wrote:
What is your CPU score from Nortin Sysinfo? Mine is 65.6. For some reason, the number is not showing on the graph from your screenshot.
Score on nortion Sysinfo is 93.1. but this is incorrect.
If use FSB 25/33 DX2/DX4/5x86 CPU on OPTI 495SX/SLC Chipset Motherboard. Norton sysinfo 6.0 or highter version outputs a different value than the actual one.
if use DX(25,33) or FSB 40 CPU(486DX2-80,486DX4-120 and etc). output correct value.
The solution is to use version 4.5 of Norton Sysinfo.
486DRx2-66's CPU Score is 127.0
feipoa wrote:
For Quake, I get 2.7 fps. I did not see your ISA scores for Quake in your bar chart, but found it pretty interesting that the DRx2-66 w/33 MHz FPU only scored 2.6 fps. But once you clock doubled your FPU, you get quite a boost percent-wise, to 3.0 fps.
For cachechk, my L2 cache is at 33.5 MB/s, which is slightly slower than your 34.7 fps. For RAM read speed, I got 16.1 MB/s, while yours is only 9.4 MB/s. Is that correct? The worst speed I've seen was around 14 MB/s. Could you report the cachechk RAM write results? e.g. cachechk -w -d -t4
my cachechk is can't run with option -d.
i ran cachechk -w -t4 , and cachechk -t4
feipoa wrote:
The best memory read times tend to come from SiS Rabbit-based boards, at 23.6 MB/s.
My Motherboard's Memory performance is so poor..
feipoa wrote:
Did you try your DRx2 at 2x40 MHz?
Appear POST Screen. and Freeze during boot.
feipoa wrote:
Based on your results, it seems like the DRx2 might be worth it if your board has VLB. In my mind, though, if a board has a 486 CPU socket, then using one of these PGA-132 CPU upgrades looses some of its prestige. Does your board have a 486 and 386 CPU socket, or just a 386 CPU socket?
I also have a DRx2-66 and would some day like to compare it against the 486SXL-40 on my Mark V Baby Screamer. I would also like to attempt 486SXL2-66 operation with lots of cooling. An SXL2 at 2x25 MHz just isn't that exciting for me.
I like your BIOS options for cache and RAM timings. I haven't seen a "Cacheable RAM address range" option in these BIOSes before. Does it go higher than 64 MB?
option only available 4M,8M,16M,32M,64M.
feipoa wrote:
Slow Refresh should be enabled for optimal performance, assuming your RAM supports this. Most do.
If Enable Slow Refresh. Speedsys score and Doom fps reduced. maybe 3~5%.
My Memory is 60ns FPM Memory
feipoa wrote:
You mentioned that using a CPU cooler improved performance. This seems very odd to me. Could you benchmark DOOM with the cooler and without the cooler? What are the two scores?
There was a slight improvement in the use of VLB graphics cards and a big improvement in the use of ISA graphics cards.
do these results seem OK ? I have a board based on the Opti 82C495XLC chipset with a TX486DLC cpu at 40mhz. The board had some cache-size jumpers tampering and some weird PTTP cache ICs. Although amibios reported 128kb it was clear that only 32kb were used:
(you can see the modified jumpers on the right)
So after messing a little and asking around if someone has the same mobo that works OK I came up with these settings (and using other SRAM chips that are known to be real)
If I put 0WS everywhere in the BIOS I can get ~28,5mb/sec writing speeds but the others stay pretty much the same more or less.
In DOOM benchmark I get ~9.7fps with a WDC vga. Wolf3D is almost 30fps at ~29.7
So my question is, does this look OK? Maybe even the 128kb jumper settings are not correct or something and mess the system? I see other results here with 386DX40 cpus that are better - is it a chipset/mobo thing? The ram I use is 8x1MB @70ns.
I would run a test to make sure that the correct amount of memory is beig cached. Sometimes there is a separate jumper that controls the size of the tag RAM (though not usually documented in an obvious way).
It looks fine to me. You should do the benchmarks with 32KB cache and compare it to 128KB cache. It should be some increase. Otherwise memory throughput will not change with more cache. It only caches more data. Memory throughput depends on wait states and chipset used.
damn I can't believe I missed these replies - thanks for the feedback!
@Anonymous Coward
what test do you suppose I can run? This is a 386 board so tag is just 8kb. I found no documentation relating to the cache - someone on fb sent me photos of the manual as well and there is nothing there - just a mention in the features section that the board has "128kb built-in high-performance write-back, direct mapped cache" . It makes sense since the cache "jumpers" aren't really jumpers - in an unaltered state the traces are not cut and the board is fixed to 128kb size.
This is what I get with the "stock 128kb of cache" , the stock AMIBIOS reports 128kb but at this point I just think it's patched to display that. MR-BIOS goes bonkers for example, PC DIAGS also just detect 32kb
both correctly display 128kb with the fixed jumper setting and other cache ics that are known to be good.
@kixs
good call maybe I solder again the jumpers and try some DOOM benchmark
I will post later a screen with 0-wait states , there is some improvement. Overall the board feels snappy but I think it's mainly because of the 486DLC cpu.
Maybe I even use 15ns cache, atm the 128k are 20nd and only the tag is 15ns.