Your jumper settings are almost certainly wrong with the Am486 CPU, can you add the motherboard vendor and model to the description? I could try helping you out with the jumper configuration to get it working right. The AMD chip shouldbe about 10-15% faster than Cyrix.
Will do. If RUclips will let me I will also post a link to some pics of the manual and updated jumper settings. Last night I couldn't resist. I put the AMD CPU back in, put the weird jumper settings back in and changed the VLB wait state jumpers to ... the other setting. Then I played various games on it for around an hour and it was fine. but turning off "Turbo" still didn't work. The only game this is a problem for is LHX. All the other games I have on the 486 run fine at full speed. I'll try and get these pics done tonight.
Greetings. I have taken pics of my current jumper settings and pages of the manual that I think are relevant and placed a link in the video description. You should be able to download the high res pics. Just took these with my phone but it has a pretty good camera. I haven't changed any of the BIOS settings. According to the manual I should just have JP28 2-3 closed. However looking through the link that Frank Delettre also here in the comments linked to, in a newer rev of the mobo it says "AM486DX4 (8B)" with totally different settings. Please see my response to him as well. I would assume the 8B refers to the 8K L1 cache version? Mine is the DX4-100V16BGC. But the manual doesn't talk about a "16B" version. The whole point of this upgrade was to upgrade from what as best as I could tell was 8K L1 in the IBM/Cyrix to the 16K L1 I know is on this AMD 486. I also haven't done anything in the BIOS. I know I have the cache enabled but that's about it. Cheers!
@@c90s_tech That explanation basically seals it, I don't really even need to look at the manual (Frank Delettre is on the money here) and I have a feeling you are screwed. Let me explain: The 8B setting that is referred to in the later revision pertains to the cache, and what matters here is the B more than the 8. The L1 cache amount on the CPU can not be changed or partially used on these motherboards unlike the L2 cache on the motherboard so what you need to set the jumpers for is to tell the Motherboard that A) It's an AMD chip at 3.3V running at 33MHz FSB, B) the L1 cache is in Write-Back mode. Your board revision's manual does not have an option to tell the BIOS about B) so it's either i) The board revision does not support WB L1 Cache (which is what I think might be going on here, WB revision chips from both Intel and AMD were out long after the original introduction) ii) The BIOS version the earlier revision shipped with does not have support for WB L1 Cache. If I am right and i) is what's going on here, you probably want to go with an Intel 486DX 16KB WT Cache or AMD Am486DX 8KB WT Cache chip instead. If I am wrong and ii) is what's going on, then you may actually need a BIOS update. Try to find a newer BIOS image on the net or ask someone with the same board but a newer revision to dump their BIOS for you BUT be sure to chek that the BIOS chips/sizes are the same and flash at your own risk, newer revision BIOSes are not necessarily backwards compatible and may brick the motherboard.
Hey thanks for the info. I think I have the jumpers proper now after the rebuild into the "ugly case". Found all kinds of little issues with this build during the transplant.
IBM manufactured processors for Cyrix, and were allowed to make some with their own branding, so IBM 486s and 5x86s are Cyrix ones. Cyrix 486 is in my experience slightly slower than Intel 486s. AMD 486s are generally 1:1 with Intel, but are available with higher clocks. Some benchmarks are not very dependent on cache, and won't notice any improvement with L2 cache or not. Others will be night and day. Getting the BIOS settings nailed in right, with fast memory timings and such can provide massive performance improvements. Finally, Descent doesn't really run properly smooth until you get to about a Pentium 100 or so. Not tried it on an AMD 5x86 running at 160MHz though, so that might do it as well.
I know this video is 3 years old but I have a curious question, all the cache chips need to be the same regardless of the brand or it can be mixed if they have the same speed?? For example a Winbond and IDT combo with 32x8 for a total of 256K
As long as everything is the same speed and capacity different brands 'should' work just fine. I've never tried it though. Thanks for the question and if this is for a build, feel free to let me know how it goes. Cheers.
@@Mr1p0d Oops I replied from my other account. Watch enough of my videos and you'll know the 486 is my favorite CPU of all time. K6-2 is my second favorite. I see you you're looking add to the cache already there. Having an Acer that old is pretty cool. Since it already had cache you may not see much of an improvement with more. Like single percentile if that. Experimentation is part of the hobby anyway. Good luck.
Hello, are you using Phil's old dos benchmark pack? In the new ones he include CacheCHK, I found very useful to test if my setup was all set correctly as I suspected my PCChips motherboard had fake L2 chips (which has) but also has a COAST module which work ok when you set the kind of memory correctly in the BIOS.
Hey I had no idea. Yes I was using his DOS Bench suite. I think the version I have is quite old since LGR I believe was using Phil's suite when he did his DX2-66 L2 Cache upgrade and that was a a little while ago. Sorry to hear about your PC Chips mobo. That sucks. Decades later and those jerks are still haunting us with their fraud! Cheers and thanks again!
Hello. When switching from the IBM (that is basically a Cyrix) to the AMD, the mainboard settings were not changed, and the BIOS keeps configuring the AMD as a Cyrix. That explains why the BIOS message shows Cyrix (Cx) and why sysinfo detects a Cyrix too (that explains the wrong clock speed, the two CPUs do not use the same instruction timing). The mainboard settings should be configured to "AM486DX4 WB" and everything should work fine.
Frank did you watch the video all the way through? Yeah I had the jumper settings totally wrong the firs time, the second time I had them correct, but the CPU was running way too slow. The 3rd time I had JP28 set with the same jumpers as a DX4-120Mhz and it was WAY faster. The POST summary always showed the clock speed at 100Mhz. All that said I don't think I tweaked the cache settings at all. The Mobo manual talked about a setting that needs to happen with L2 cache installed. but I didn't make the changes because it SEEMED like it was talking about a single bank L2 cache (plus tag chip of course). I have the AMD DX4-100 back in the mobo on the weird jumpers, and I changed the wait state for VLB. The system is a lot more stable now. But "turbo" still doesn't change when I turn it off. I'll take a closer look at the cache settings tonight after work! Thanks for the feedback!
@@c90s_tech You're welcome. I think the settings are still not good for your CPU (if they are as visible at 14:33), those are for an Am486DX4 and not for an Am486DX4-WB (that is what your CPU is). JP27/28/29/90 should respectively be set to 1 & 2, 3 & 4 / 1 & 2, 4 & 5 / 1 & 2, 4 & 5 / 3 & 4, 5 & 6 to reflect the L1 WB. (source : stason.org/TULARC/pc/motherboards/A/AMPTRON-INTERNATIONAL-INC-486-DX-6900-VER-1-75.html. Settings are given for a 8KB WB L1, but size does not matter, for the system, only write policy does).
Hey thanks for the info and the link. What I understand this is a "BOG Standard" Amptron board. They were super common 'back in the day' it seems. I had a 486 back then of course but that was over 20 years ago and I couldn't tell you at all what mobo I had. I have updated the description of the video with a link to pics to the actual manual that it came with. The link you provided seems to be the same jumper settings and the layout of the board looks the same except the manual you linked is for a mobo that supports the Am5x86 133Mhz, and the mobo I have tops out at the DX4 120Mhz. The manual just says that only JP28 2-3 should be closed. But the massive speed boost I got I have I have jumpers 1-2 & 4-5 closed like it says to do on the DX4-120. The video is now in full 4K finally BTW. Cheers!
Ok, here is my thought on your 486 video. FUN! Seriously, sometimes we need to see a failure (I just posted one with my clock chip failure) and enjoy the fellowship in it. Nice set up and I think your editing and voice were on point for the content. I feel like you are a worthy sub my friend!
@@c90s_tech I SORT of mentioned you in my latest video. I said I was talking with a fellow small channel owner about how many 486 videos are out there. You did say that and I agreed. Love to do a collab someday :)
Yeah something it up with RUclips. This is a 4K upload. Uploaded it last night, it did it's initial processing and never got past 360p. I'm rendering a new video and will be uploading it shortly. But it's a 9GB file and will take a couple hours.
Often the limiting factor is the graphics card, when benchmarking with Phils benchmark suite. I'm not surprised, that you aren't getting a linear perfomance gain. I myself saw the biggest performance differences testing different video adapters on the same system. Please keep going with these videos, I've enjoyed it very much. Btw. you've got a new subscriber :)
Thanks for the insight. I honestly didn't think swapping out the Trident for the Cirrus Logic (per your suggestion (mostly) actually) made a difference. I think it's a lot of fun to take something that's so limited like a 486 and try to pump it up as fast as it will go, more cache, faster components, but the CPU is still running at it's stock speeds. Thanks for the suggestions and the sub. Cheers!
Your jumper settings are almost certainly wrong with the Am486 CPU, can you add the motherboard vendor and model to the description? I could try helping you out with the jumper configuration to get it working right. The AMD chip shouldbe about 10-15% faster than Cyrix.
Will do. If RUclips will let me I will also post a link to some pics of the manual and updated jumper settings. Last night I couldn't resist. I put the AMD CPU back in, put the weird jumper settings back in and changed the VLB wait state jumpers to ... the other setting. Then I played various games on it for around an hour and it was fine. but turning off "Turbo" still didn't work. The only game this is a problem for is LHX. All the other games I have on the 486 run fine at full speed. I'll try and get these pics done tonight.
Greetings. I have taken pics of my current jumper settings and pages of the manual that I think are relevant and placed a link in the video description. You should be able to download the high res pics. Just took these with my phone but it has a pretty good camera. I haven't changed any of the BIOS settings. According to the manual I should just have JP28 2-3 closed. However looking through the link that Frank Delettre also here in the comments linked to, in a newer rev of the mobo it says "AM486DX4 (8B)" with totally different settings. Please see my response to him as well. I would assume the 8B refers to the 8K L1 cache version? Mine is the DX4-100V16BGC. But the manual doesn't talk about a "16B" version. The whole point of this upgrade was to upgrade from what as best as I could tell was 8K L1 in the IBM/Cyrix to the 16K L1 I know is on this AMD 486. I also haven't done anything in the BIOS. I know I have the cache enabled but that's about it. Cheers!
@@c90s_tech That explanation basically seals it, I don't really even need to look at the manual (Frank Delettre is on the money here) and I have a feeling you are screwed. Let me explain:
The 8B setting that is referred to in the later revision pertains to the cache, and what matters here is the B more than the 8. The L1 cache amount on the CPU can not be changed or partially used on these motherboards unlike the L2 cache on the motherboard so what you need to set the jumpers for is to tell the Motherboard that A) It's an AMD chip at 3.3V running at 33MHz FSB, B) the L1 cache is in Write-Back mode. Your board revision's manual does not have an option to tell the BIOS about B) so it's either i) The board revision does not support WB L1 Cache (which is what I think might be going on here, WB revision chips from both Intel and AMD were out long after the original introduction) ii) The BIOS version the earlier revision shipped with does not have support for WB L1 Cache.
If I am right and i) is what's going on here, you probably want to go with an Intel 486DX 16KB WT Cache or AMD Am486DX 8KB WT Cache chip instead. If I am wrong and ii) is what's going on, then you may actually need a BIOS update. Try to find a newer BIOS image on the net or ask someone with the same board but a newer revision to dump their BIOS for you BUT be sure to chek that the BIOS chips/sizes are the same and flash at your own risk, newer revision BIOSes are not necessarily backwards compatible and may brick the motherboard.
JP28 has 2 jumpers, 1-2 and 4-5; it should only have 2-3 shorted. (For the AM 486 DX4-100.
Hey thanks for the info. I think I have the jumpers proper now after the rebuild into the "ugly case". Found all kinds of little issues with this build during the transplant.
That was a fun video. I had a 486dx4-75 that I ran Descent, Doom, and even Truespace 2 on. It was a fun CPU.
IBM manufactured processors for Cyrix, and were allowed to make some with their own branding, so IBM 486s and 5x86s are Cyrix ones. Cyrix 486 is in my experience slightly slower than Intel 486s. AMD 486s are generally 1:1 with Intel, but are available with higher clocks.
Some benchmarks are not very dependent on cache, and won't notice any improvement with L2 cache or not. Others will be night and day. Getting the BIOS settings nailed in right, with fast memory timings and such can provide massive performance improvements.
Finally, Descent doesn't really run properly smooth until you get to about a Pentium 100 or so. Not tried it on an AMD 5x86 running at 160MHz though, so that might do it as well.
I know this video is 3 years old but I have a curious question, all the cache chips need to be the same regardless of the brand or it can be mixed if they have the same speed?? For example a Winbond and IDT combo with 32x8 for a total of 256K
As long as everything is the same speed and capacity different brands 'should' work just fine. I've never tried it though. Thanks for the question and if this is for a build, feel free to let me know how it goes. Cheers.
@@KitelessRex Actually is for an Acer prebuilt 486 machine that came with only 128k of cache
@@Mr1p0d Oops I replied from my other account. Watch enough of my videos and you'll know the 486 is my favorite CPU of all time. K6-2 is my second favorite. I see you you're looking add to the cache already there. Having an Acer that old is pretty cool. Since it already had cache you may not see much of an improvement with more. Like single percentile if that. Experimentation is part of the hobby anyway. Good luck.
@@c90s_tech I confirm it, it works when the chips are from the same timing and different brands, now the Acer has 256k of cache meaning more RAM later
Hello, are you using Phil's old dos benchmark pack?
In the new ones he include CacheCHK, I found very useful to test if my setup was all set correctly as I suspected my PCChips motherboard had fake L2 chips (which has) but also has a COAST module which work ok when you set the kind of memory correctly in the BIOS.
Hey I had no idea. Yes I was using his DOS Bench suite. I think the version I have is quite old since LGR I believe was using Phil's suite when he did his DX2-66 L2 Cache upgrade and that was a a little while ago. Sorry to hear about your PC Chips mobo. That sucks. Decades later and those jerks are still haunting us with their fraud! Cheers and thanks again!
Hello. When switching from the IBM (that is basically a Cyrix) to the AMD, the mainboard settings were not changed, and the BIOS keeps configuring the AMD as a Cyrix. That explains why the BIOS message shows Cyrix (Cx) and why sysinfo detects a Cyrix too (that explains the wrong clock speed, the two CPUs do not use the same instruction timing). The mainboard settings should be configured to "AM486DX4 WB" and everything should work fine.
Frank did you watch the video all the way through? Yeah I had the jumper settings totally wrong the firs time, the second time I had them correct, but the CPU was running way too slow. The 3rd time I had JP28 set with the same jumpers as a DX4-120Mhz and it was WAY faster. The POST summary always showed the clock speed at 100Mhz. All that said I don't think I tweaked the cache settings at all. The Mobo manual talked about a setting that needs to happen with L2 cache installed. but I didn't make the changes because it SEEMED like it was talking about a single bank L2 cache (plus tag chip of course). I have the AMD DX4-100 back in the mobo on the weird jumpers, and I changed the wait state for VLB. The system is a lot more stable now. But "turbo" still doesn't change when I turn it off. I'll take a closer look at the cache settings tonight after work! Thanks for the feedback!
@@c90s_tech You're welcome. I think the settings are still not good for your CPU (if they are as visible at 14:33), those are for an Am486DX4 and not for an Am486DX4-WB (that is what your CPU is). JP27/28/29/90 should respectively be set to 1 & 2, 3 & 4
/ 1 & 2, 4 & 5 / 1 & 2, 4 & 5 / 3 & 4, 5 & 6 to reflect the L1 WB.
(source : stason.org/TULARC/pc/motherboards/A/AMPTRON-INTERNATIONAL-INC-486-DX-6900-VER-1-75.html. Settings are given for a 8KB WB L1, but size does not matter, for the system, only write policy does).
Hey thanks for the info and the link. What I understand this is a "BOG Standard" Amptron board. They were super common 'back in the day' it seems. I had a 486 back then of course but that was over 20 years ago and I couldn't tell you at all what mobo I had. I have updated the description of the video with a link to pics to the actual manual that it came with. The link you provided seems to be the same jumper settings and the layout of the board looks the same except the manual you linked is for a mobo that supports the Am5x86 133Mhz, and the mobo I have tops out at the DX4 120Mhz. The manual just says that only JP28 2-3 should be closed. But the massive speed boost I got I have I have jumpers 1-2 & 4-5 closed like it says to do on the DX4-120. The video is now in full 4K finally BTW. Cheers!
@@c90s_tech It seems that your mainboard does support CPUs with L1 WB, if I refer to 13:38. And the IBM uses a WB L1 too anyway.
Awesome Vid! and yes how do you capture the 486 system? I can't wait for the next one!
That's one vote for the video capture overview. Really all I needed. haha. I'll put it on the project list. Also thanks for the kind words!
Ok, here is my thought on your 486 video.
FUN!
Seriously, sometimes we need to see a failure (I just posted one with my clock chip failure) and enjoy the fellowship in it.
Nice set up and I think your editing and voice were on point for the content.
I feel like you are a worthy sub my friend!
Thanks so much for the feedback and the kind words, very VERY much appreciated!
@@c90s_tech I SORT of mentioned you in my latest video. I said I was talking with a fellow small channel owner about how many 486 videos are out there. You did say that and I agreed. Love to do a collab someday :)
Old tech is fun
360p?
Yeah something it up with RUclips. This is a 4K upload. Uploaded it last night, it did it's initial processing and never got past 360p. I'm rendering a new video and will be uploading it shortly. But it's a 9GB file and will take a couple hours.
Oh sure noooooow it's finally starting to process to 4K. Oh well. Only took like 48 hours. :D
lucky... on my last metro redux gameplay video, process is %100 and resolution still 360p with 4 days. also, video is almost 100 gb
Often the limiting factor is the graphics card, when benchmarking with Phils benchmark suite. I'm not surprised, that you aren't getting a linear perfomance gain. I myself saw the biggest performance differences testing different video adapters on the same system.
Please keep going with these videos, I've enjoyed it very much. Btw. you've got a new subscriber :)
Thanks for the insight. I honestly didn't think swapping out the Trident for the Cirrus Logic (per your suggestion (mostly) actually) made a difference. I think it's a lot of fun to take something that's so limited like a 486 and try to pump it up as fast as it will go, more cache, faster components, but the CPU is still running at it's stock speeds. Thanks for the suggestions and the sub. Cheers!
Is this a 100 sub special? ‘Cause 100mhz?
Haha odd timing I guess. Shame it's stuck at 360p I'm trying to fix it by rendering (2 hours) and uploading (3 hours) a replacement video.
c:\90s_tech doesn’t bother me.