Too nice to be true? Intel i440LX was upset about my plans.
Вставка
- Опубліковано 25 кві 2024
- In this video I'll repair a Socket 370 mainboard MSI MS-6159, get Creative Sound Blaster fixed and try to get get some unsupported Coppermine CPUs running. Let's see if it worked.
BIOS Patcher:
• BIOS Patcher: unleashi...
FastVid by @philscomputerlab
• Intel DOS Graphics Boo...
Patreon:
/ necroware - Наука та технологія
PCI ven 4348 dev 4447 Did you have some crappy PCI post code card inserted by any chance ? Edit: just saw the post card and... YES it's that crappy PCI post code card. It actually identifies as a pci card. Edit2 : BTW the isa part will never work.
Yes, the ISA part is disconnected :) That is indeed a crappy POST card, but for PCI it works fine. The strange device had nothing to do with that though. After I finished the video I found out, that it was a strange behavior triggered by the USB mouse. Still don't understand, how that could be shown as PCI device, but probably that is some confusing incompatibility.
This is the post code card without the chip on imgur W6VIvud . As you can see the isa part isn't even connected, the traces just run to vias underneath the chip and end there. edit, If I recall correctly it's also a weird post code card on the PCI bus because it doesn't even use all of the pci data & address signals so in some cases it will actually give you false codes. since I can't post links I've adjusted it to just the imgur image ID
@@necro_ware easy enough to test it again, with either the post code card or the usb mouse. My bet is still on the post code card because I have the same one and I've seen that it identifies itself on the PCI bus.
@@necro_ware The chip on the post code card is actually made by WCH, vendor ID 4348
Yeah, where you say it. You are probably right. I think, as I pulled the mouse, the POST card was also not inserted, so I probably ran into wrong conclusion. I'll test it later, but this is a good point. Thank you!
These Celerons were overclocking monsters. I used to have a 667Mhz one that easily ran at 1050Mhz. Probably even higher, but I think I reached the limits of the mainboard I was using back than.
L2 cache matters & celerons were lacking badly.
...and easily beaten by a Duron 900.
Is that similar to durex@@yx8074
I remember having such a 440LX board in my first computer and also struggling a lot to find the right settings for my nvidia tnt2 vanta. In the end, after trying numerous settings, drivers, etc, I think the fix was to set the agp speed to 1x, if I remember correctly, or maybe another bios setting that related to the agp port. This fixed the issues completely.
In the end, the issue is related to the 440lx chipset and it not being able to work correctly with agp graphics cards.
Had you tried pulling up those two pins instead of pulling them down? Another thing to tinker around here would be to play with AGP window and "GART driver"/settings. Main differences between AGP and PCI (leaving alone speed) were (a) abilities to use side-band addressing and so-called "Fast Writes" - both of which were well known to case all kind of issues, especially on Via and Ali chipsets and (b) use GART to allocate part of CPU Dram as a VRAM for textures storage. Any time you see problems with AGP while PCI works fine it smells like some kind of incompatibility (either software or hardware) w.r.t. those two features mentioned.
I've learned very much the same few years back. I got an Abit ZM6 board with 440ZX chipset that is originally used for ppga mendocino celerons. Not wanting to mod the board (other than replace leaky crappy caps), I've searched for oldest revision of coppermine celeron533A cpu. Given the fact that the CPU is 8x66 MHz and 440ZX chipset is more capable than LX (ZX is basically stripped down BX chipset), I was able to overclock it easily to 8x 100 MHz (default voltage), from the bios using good old abit softmenu. The true celeron 800 doesn't boot here at all. This OCed older revision does. Anyway it is a fine board, it has few ISA slots, SB-link, AGP works okay. I've read somewhere that 440LX has some earlier buggy implementation of AGP. Not that common and yes this can be indeed with combination of such modern cpu 😊
Great investigation! Its true rabbit hole journey. Another solid confirmation, that LX - is not good chipset for retroPC, at least if you have only one retroPC instead of few different types and generations. Waiting for the quest continuation!
I always learn so much from your videos!
Interesting little board. I'd do Voodoo2 SLI with the onboard Rage as the 2D card.
The AudioPCI was great for one thing... full duplex support. You could play and record at the same time, so you could use a program like Cool Edit Pro to do multi track recording back in the 90's when that equipment was absurdly expensive.
it had good 3d audio support too. it is a solid sound card for windows.
@@sakitoshi and also large Sound Font support before it was branded Sound Font.
Great Video 👍
Finally another video
Thank you so much, really fun to watch your videos
Awesome video and enjoyed going down the rabbit hole...😊
Being that I think I have only seen LX board and seen many BX boards
Again thank you for insight and knowledge
440lx and bx don't support 256mbit chips, so that's probably why only half the amount showed up. Some bx boards support up to 1gb with 4 slots. But then the modules have to be 16x128mbit double sided. ✌️
NW thank you so much to your awesome videos ! You are such an inspiration
Thanks so much for the great content!
I read that despite being a cut down LX, the EX chipset works with SSE.
A socket 370 EX board would be quite rare tho I think, I always got the feeling that all celeron LX boards were made to "dispose" warehouses full of surplus LX chipsets after the BX came out.
i remember that those intel and other BGA chips are assemblied former my company and i ve seen that huge amounf of wafers and final chips for shipping accordingly. i was very busy to handling those Quality assurance at that time around 1996~2000.
Interesting motherboard, is the first one of this kind (with integrated graphic) of that era that i see! great work as always!
ha. Thankfully you dont use some Chips&Technologies based socket7 boards. It was a nightmare 😂 - slow as hell 😁
And on branded (like HP/Dell) PC I have seen integrated video at least on 486 VLB era.
@@VShuricK never came across one, now im will be on the look for one, at least for trying them out. I have some socket 7 waiting for me to repair them an tinker, tho not having the time. its a shame! all that suffering waiting for me! hahaha
For unknown PCI slots i usually use some period correct or little newer live Linux distro. Knoppix 3.7 could be in your case. Check what lspci utility reports ...
I forget which system I had with a PCI128 on board, if you like your midi, best with the 8Mb GM bank that came with the Ensoniq, I think Creative only supplied the 2Mb and 4Mb options - it wasn't quite software wavetable, and it wasn't quite hardware wavetable, well I guess it was hardware wavetable but with the memory borrowed across the PCI bus
I'm pretty sure the MSI board was originally designed for OEM use for a value oriented line of Compaq Presario PCs and later sold standalone as an cheap budget board. It would make sense as the Compaq machines were only sold with up to a Celeron 533 Mhz CPU in them which would have been Mendocino parts. My sister's ex husband had one of those Compaq machines with this board in it way back when and it was awful! I can remember working inside of it and aside from upgrading the RAM or HDD there really wasn't anything else that could be done with it for more performance.
By the time the socket 370 Celerons were released the successor to the LX, the BX and newer still ZX chipsets were out as mainstream parts with the older LX chipset being relegated to ultra budget machines due to it's age. The Intel 8xx series chipsets then came later in 1999 with the release of the P3s that used the 133MHz FSB.
Also, a lot of LX boards that had AGP slots tended to have issues delivering enough power to the AGP slot. This is due
Many LX boards only provide 2A of power on the 3.3 Volt rail which complies with the AGP 1.0 spec (AGP 1x & 2x data speeds), but most AGP 2.0 (AGP 4x data speed) compliant cards want 6A at 3.3 Volts. The 3dfx Velocity 100 (which is a card I also had back then!) acts like a AGP 1.0 card voltage wise (because it's really a PCI card in an AGP slot.) Most Nvidia and ATI cards made after ~1999 are AGP 2.0 compliant and want more power. Hence the crashing when enabling 3D mode on them.
For best results with any graphics card that allows for AGP 4x or 8x data speeds on socket 370 needs to be in one of the later Intel 8xx or Via Apollo Pro 133A chipsets. Some BX chipset boards deliver enough power for those cards to run properly but not all. I had an Abit BH6 BX chipset board with an overclocked Celeron 300A in it that worked fine with a 3dfx Velocity 100 and a Nvidia TNT2 M64 card but would not work with the ATI Radeon DDR (R100 based) card I bought in the fall of 2000. However another machine I had with an Asus P3V4X mainboard with a Via Apollo Pro 133A chipset and a Celeron 566A overclocked to 875 MHz in a slotket adapter worked fine with the Radeon DDR.
yes I had that aopen MX3L +533A (mendocino) +TNT2 M64
WOW! I have very similar MB, but Slot1 + ES1371 sound😁
I got one that's slot 1 butt, Trident 4dWave-DX chip🙃
As far as I remember, the 440LX had a somewhat buggy AGP implementation. Back in the day I had a 440LX system which was horribly unstable with both ATI Rage 128 and Nvidia TNT2 but worked perfectly with a Voodoo Banshee. I used a Celeron Mendocino CPU, by the way.
No, you are wrong. I have a 440LX system right now with a Matrox G200A card and it's perfectly stable.
Not my case, it was really stable.
Back in the days I had a cheap amptron 440lx mobo with a TNT2 m64 32mb agp card, it never had any issues, but i only had a 366 celeron cpu with that setup.
the LX series of chipsets also usually had poor agp current supply implementations at a time when 3d cards were starting to want more power from the agp slot (above the 6A it would provide, leading to yet another source of instability).
My first relatively up-to-date PC in 1999 had 440LX with Riva128ZX and worked quite stable if my memory serves me well.
I have one working PIII-class system which is running a Coppermine Celeron 700 at 528MHz since the clock divider doesn't support it. It still reports it as running at 700MHz though.
I would just stick to the earlier chips with the LX chipset. Personally, I always thought the LX chipset was too limited, and at the time I thought I was being clever buying a VIA chipset S370 board, but much later I heard VIA chipsets were poor performers. I never ran into performance issues, aside from with my first celery, but I just put that down to it being a celery. That thing didn't even have enough processing power to decode a DVD! In the end I bought a cheap MPEG2 decoder card, which actually worked out great since I could use the TV output on the card to make the PC into a DVD player, so we could watch movies on the big (well, bigger) screen.
there's some vogons and tomshardware threads on the voltage regulators for certain 440lx agp slots. seems some cards in 3d mode pull more in the 2x slots than they can handle, and it causes crashes. basically, its pulling more than the roughly 6A those can handle, and boom. the 4x slots provide more current. and pci cards don't tend to go over that 6A, as the pci spec is around 25w (or 5A) if i recall.
not sure if this is the case here, but its definitely a possible.
if you have an agp card with an external power port, you can try that. if it runs, those motherboards have the current supply issue that was noted. the vogons thread called 'early 440lx agp voltage regulator issues, anyone tried cards with additional power connector' has a workaround for asus p2l97 boards, which may potentially give a direction for a fix, if those two do indeed have that current supply issue. (i'd link the thread, but can't, for obvious yt reasons)
cheers.
Yes, I considered this as well, but then those AGP cards wouldn't work with Mendocino as well, but they work absolutely stable there.
Тут как я понял, речь идёт о недостатке питания для видеокарт через agp слот. Бывало решали проблему прокидыванием перемычки с разъема питания. Но там надо смотреть, как само питание слота организовано. Если с отдельного стабилизатора, то придется встроенный отключать
My ABit BH6 rev 1.01 (or similar, early revision not supposed to support coppermine) ran my Pentium 3 700 @ 933 mHz rock solid for years in a slotket by just cranking the bus to 133 and upping the Vcore a bit. None of my Riva TNT, TNT2 Ultra, GeForce 256 cards had any problems running at 2/3x133 bus speed either. Overclocking used to be such a joy... Also it used to save a LOT of money, which is one of the reasons it's pretty much dead now =D
16:50 i was probably THAT lucky having accidentally bought an early revision Pentium III 700 and a great silicon quality one as well.
It might have been because of the Slotket i used as well, it did have jumpers for some stuff, one of them might have been the reset mod.
EDIT: i think this was one of the major reasons i didn't buy a slot 1 p3 700, i think i needed to use a Socket 370 model in a slotket to even have a chance of it working. Still, the BH6 revision i had wasn't supposed to be able to work with a coppermine cpu.
Anyway, this was A LONG TIME AGO, it's hard to remember all the specifics.
Great video!!! Could you please make with Tualatin on 440bx?
i remember those mods
ppga>fcpga
fcpga>>fcpga2
v mods
bus mods for via chipsets
socket 370 is probably the most moded processor socket i ever moded :]
mostly fcpga to fcpga2 and via fsb mod i have no experience with lx chipsets though , the oldest p2/p3 board i had was bx or via apollo
with the sbemu driver you can get good fm emulation in dos with the sb128
ATI Rage Pro could perhaps be supported in Amithlon (ATI Kernel) ...the soundcard definitely works with Amithlon. Nice!
This would also work out fine with a Voodoo 4 or 5 card as they were also Pci cards with an agp disguise
I got the same probs with celerons above 600 and slotkets. Even if i use tualatin slotkets! 😢
From being a fan of ATI hardware back in the day I recall that the drivers were always buggy and problematic. They made great hardware and things like tv tuners with gfx accelerators, and then dropped the ball on driver stability and performance really bad.
Great prices, great hardware, on paper great performance. Sometimes a specific driver for the right game or two and you'd get that great performance in that until the next update to fix a feature that was broken in that one version.
Considering ATIs history a lot of that hardware probably went end of life without ever having a single good driver.
Vendor ID 4348 / Product ID 4447 = WCH WinChipHead PCI USB Controller. Manufacturer appears to be seriously old-school MICRELEC. Likely an early USB2.0 device of some sort on the mobo. Running a (older) 32-bit Linux distro will likely tell you more useful info. I believe I have an old ABIT Socket-370 in my collection with this same device on it -- the MICRELEC name sounds familiar to me.
Thank you so much. That PCI bus cause much funny if it comes to gaming video cards. Cause that ferra possibly make it busy slightly over than 100 percent while IDE and some other hardware cannot exchange with northern bridge and CPU as it had been till i815 and so on. Commonly, VIA Apollo Pro 133 and also 440BX are the best for its class and the age, unluckily I cannot get 512 mb SDRAM double-side to accomodate it
For my SIS P4 board, in W98, i had to install a specific AGP drivers, you could try searching for such drivers for those boards.
for 440lx win98se have inbox agp drivers.
Just a reminder about AGP: The whole point of the technology was to be able to modify video data "as if it was main memory" so the graphics driver code does not need to copy the data back and forth BUT it requires an updated memory manager as video memory prudently "appears" in main memory space.
So, could Windows issues be related to a clash between the Windows memory manager and the video driver ?
The video driver not being granted access to a certain memory range required for 3D functions.
It could be of course, but the question is more about the difference between mendocino and coppermine cores. Otherwise the same AGP cards wouldn't work with both types of the CPUs, however with mendocino everything works like charm and the issues appear only when using the coppermine CPU.
I have a 1GHz Coppermine PIII, and I believe they went up to 1.13GHz, although at 1.13GHz they were unstable and were recalled before being re-released at 1.1GHz and 1.13GHz when Intel had fixed the issues. The Tualatin came at that point.
Are you use 1Ghz in LX board? 66x10 ?
Just for curiosity, maybe I missed, but the issues with the AGP cards still affect old Mendocino CPUs? Also, I remember being in BIOS some AGP related settings (like AGP fast-write or similar, if I remember correctly); try to see if those are changing the behaviour. Maybe there are options set and hidden in the BIOS? That specific setting (AGP fast-write) I remember gaving me a lot of headache in some motherboards with Ati Radeon 9250 AGP, random freezing on some games, BSOD on XP SP3 and Win98SE.
About that mistery PCI device, seems to be from a vendor (4348) that made mostly serial/parallel I/O adapters for PCI and USB, so, maybe that particular mouse really had something to do with it; it might be a serial mouse in disguise (LOL), however, another mouse did the same?
I believe that the fast-write feature was a later addition to the AGP standard and wasn't used with AGP 2x which 440LX supports. There's a side-band addressing which he may be able to disable and I would also try to run it in 1x mode just to try improving the stability.
@@petrkubena yes the AGP fastwrite has problem with SSE support CPU..as i know
I thought it was well known that 440LX has some AGP incompatibilities. I have a few 440LX boards and I have tried probably 15 different AGP cards and they mostly all work just fine. Matrox G200/G400, TNT/TNT2, banshee/Voodoo3, Savage 3D/4, Rage 128/128pro and yes even the Rage pro but the Geforce cards do not work. I haven't used too many newer cards, maybe the newer Geforce and ATI cards don't work also. Tempted to try some Radeons or newer Geforce cards in it, maybe Savage 2000 or Rage fury maxx even. Also, the Rage pro drivers are a mess. I remember needing multiple different driver versions to play many games so I wouldn't be surprised if that might be the reason for the Rage pro not working
The actual problem is, that those AGP cards don't work with Coppermine CPUs, but with Mendocino they work just fine. At least from what I tested. So there is no issue between LX and AGP, it appears first with Coppermine CPUs.
@@necro_ware Yeah, I understand that part. Just that the problems with the Rage pro might be driver related? Also the AGP Geforce FX works in the LX board with the Mendocino CPUs? My LX boards don't work with the Geforce cards I have tried but that was only Geforce 256 (SDR and DDR) and my Geforce 2 pro
@@necro_ware it could be driver being aware of incompatibilities of 440LX chipset and activating some workaround depending on CPUID (i.e. it checks for the CPU and not for chipset)
@@necro_ware I was able to duplicate your findings with my own LX board and ATi Rage pro. Although I used a Katmai Pentium 3 CPU. The board I'm using is from an IBM intellistation m pro 6898. I noticed loading powerstrip caused a BSOD when the Pentium 3 was in the system also. Everything worked fine with a Pentium 2 in the system and I also tried both CPUs and the Rage pro in my BX board (Aopen AX6BC)
20:45 - не смотря на то что это по факту PCI адаптер, не стоит забывать, что работает этот слот на 66МГц. Даже есть проект одного блоггера, где используя AGP ➡️PCI переходник, он использует серверные PCI Ethernet карты на 66МГЦ PCI и получает ускорение передачи данных в сравнении с обычными 33МГц PCI слотами.
Yeesh, i also have that 766MHz celeron with 66MHz bus! The fastest (highest multiplier) official CPU on 66MHz bus by intel.
Might be able to uninstall the AGP gart driver and force the AGP device into pci compatibilty mode. Might be a bit tricky for intel chipset as Widows would come with the driver. Might be possible to use update driver in device manager to force the AGP bus to use a generic PCI-PCI bridge driver instead of the proper agp bus drivers. AGP Aperture size setting in the bios might do something
Nvidia can do it right from the drivers using coolbit registry hack to unlock agp settings and overclock page :) Maybe it is only needed to disable SBA, fastwrites or both....
At least on my Asus P2L97, Coppermine doesn't have the same problems so the issue could be purely socket related, not chipset. While seen as a Pentium II, the voltage regulator supplies down to 1.7V for my PIII 650 (rated 1.65V) and everything works the same as with my Celeron 300A/333. I don't have the same AGP power limits as with my SE440BX-2 either, the only AGP cards of mine that don't work are due to their connector.
That is an interesting information. The single difference between P2L97 and the MSI board which I tested is the CPU slot. Well voltage issues should not be relevant I guess. I have here another 440LX Slot1 board. Will have to check that out.
Ohh man!!!! This is my childhood s.cking 20-23 years ago :D yees that aopen MX3L board with 440LX...That SSE problem i know minimum 10-15 years ago. The aopen can handle the B and maybe C revision coopemine celerons till 600MHz or more with 1.14 bios rev. But the So LX and newer CPU with SSE.. the 3D applications has conflict. Can't handle well. The LX from 1998 was the very first to follow P1 MMX and not designed to coppermine and extra SSE.. ZX/BX works well with SSE....The SSE disable is an very good option.
I just so happen to have a Asus P2L97 Rev. 1.05 Slot 1 mainboard that features the 440LX chipset. I have done plenty of research that has told me that this board should support A 766mhz Celeron with a slocket adapter and the latest beta bios. Interestingly the agp slot on this motherboard only supports up to a Nvidia Riva 128 as the port does not supply enough power for newer cards to run stabile or even post if new enough.
the vogons thread called 'early 440lx agp voltage regulator issues, anyone tried cards with additional power connector' has a workaround for asus p2l97 boards linked at the bottom of it, from an old asus technical support post, if you're handy with a soldering iron.
for that board revision and rivas, specifically, but i'tll help with any other gpu that pulls too much current from the slot, i'd imagine.
Well I would recommend 440bx instead for retro build. Less problems, more stable. Asus p2b or p3b are great choice. It was still nice to see what 440lx can eventually capable of.
I think I have a PCI Rage Lt Pro kicking around some where.
Need a 300A in that bad boy!
Ever looked into graphite pads instead of thermal paste? Seems like it'd be really useful for testing here since you wouldn't have to constantly wipe off and reapply paste.
Nice, which glue did you use?
I used to run a p3 650 and voodoo 4 4500 on a 440lx board. 112 mhz fsb max if i remember right. Might have been a soyo board i think. This makes me want to get my old BP6 and other 440bx boards going again lol
Very interesting issue. My thoughts: maybe use pci to agp give to you more information about this bug.
This exact blower blew up in my face, the motor fried its windings. I'd be wary of its quality.
Oh, ok I'll keep that in mind. At least it has 2 years warranty, so I can return it if it breaks. But probably some eyes protection is a good idea then.
If I remember correctly, AGP uses a PCI communication port for things and this is probably not working correctly.
Strange. I had friends that ran pentium3 with ati cards back in the day, with win98. Though, most of us interested in gaming ran amd k6 II/III or athlon systems. And I'm not sure what agp features any of the cards used.
hi
I have modded my ASUS MEL-B to working with coppermine, but it is extreme instable, if i use AGP card.
You have similar result.
agp to high frequency?
missing agp to pci bridge driver?
It's quite unusual to see 440LX chipset with socket 360. They used to be Slot 1 boards, as far as I remember.
I have an AT form factor Gigabyte-GLA7 board with Socket 370
you may discount the soundblaster pci128 for having poor msdos support, but it is a solid sound card for windows with good 3d sound. and if you plan to run a pentium 3 machine it is to play windows games rather than msdos games. and nothing is stopping you from installing an ISA soundcard to cover msdos gaming anyway.
i have a p3 933 with a soundblaster pci128 for windows and midi with its wavetable synth and a soundblaster 16 for msdos fm sound.
Yeah, for windows it's ok indeed, software wavetable was quite decent, but I would rather go with SB Live! instead. The PCI 128 was Creative's first try dive into PCI world and that wasn't even an in house development. Also 3D audio in that card was kind of poor in my opinion. Once again SB Live! with EAX or may be Aureal with A3D is a better choice for a retro PC. Especially SB Live! today is also very cheap and easy to get.
Intel Chipset Drivers? Can't remember if this LX chipset needed them for win9xMe... I don't remember having this kinda trouble as long as agp or chipset drivers were installed.
Win98se has one, but I tried different intel drivers. All the same.
Maybe the chipset datasheets will say something.
PS: there is some stuff on the Wiki
These boards probably are intended for celerons with a 66 mhz bus only. Pretty sad boards, unless that's what you want, then it's pretty cool. I like playing with older Intel stuff, since I had AMD growing up, and through most of my adult life too.
Did you try swapping the 8 mb (4x2Mb) SGRAM memory chips of the Velocity 100 against 16 mb memory (4x4 mb)? There exists a model of Voodoo 3 2000 with only 4 chips soldered on instead of 8. I have myself this board (Velocity 100) and was wondering is it worth a try. You can get the necessary amount of 4 mb SGRAM chips from ATI Rage 128 boards.
No, I didn't try that, but that seems to be an interesting idea. I'll put it on my list if you don't mind.
@@necro_ware I would gladly wait for the results (moreso if that would be a success)!
I like my setup CUV4X-M, Coopermine 1000MHz PIII. I love PIII :-)
Used a 1.1 GHz 100 MHz FSB PIII on a CUVBX-L until 2009. Good times. Good memories.
@@kyoudaiken Tualantin? only coopermine will take me. Gold times :-)
5:52 - what exactly did you enabled in that phils program to get such boost? only linear frame buffer, or something else?
That's simple, answered everything with yes ;)
@@necro_ware but there is alert, I should not enable some write posting, because I have 440BX chipset.
I mean, the first option.
Why there is alert, I should not enable it on 440BX mobo?
Just by pure curiosity might be worth to try Windows XP and check if the graphics cards work better, Windows98 was a pain with the GPU drivers and when they changed in XP the driver model it solved so many problems that might solve these.
I remember the Coppermine + i440LX combination preventing Windows 2000 and XP from booting at all. I believe that SSE being available, but broken is what causes it to crash. SSE-aware Linux kernels will also throw a kernel panic during boot. I was able to get a kernel to boot by compiling it for Pentium Pro/II optimizations explicitly, avoiding PIII and higher which would bake in SSE.
Maybe you can try an Intel i740 Card. Intel made that card to push AGP together with the 440LX. If on hand use Riva TNT for AGP and a Vanta Chip for PCI. If you need another PCI based ATI Card give me a call.
Did you try to run a dedicated Video Card on the first board to see if that solved the problem ?
supposedly "delayed transactions" is a key feature that prevents coppermine CPUs from working with AGP @440LX m/b.
May be, yes. Any idea how I could approach this?
failing caps sometimes start to get higher capacitance so the caps you desoldered might actually be degaded since 1817uf is way higher than 1500uf
Yeah, I saw that too - higher capacitance, high ESR and high VLOSS. Which is why it's better to replace them anyway.
yes due to leakage
Always knew 66Mhz FSB boards were not a good fit for Coppermine (100/133Mhz FSB)… even without knowing about the reset signal.
My gut said it was crap ATI drivers until you identified the other cards and PCI...
agp fast write i think is the issue
I think the voltage regulator cant supply enough power to both the coppermine cpu and the graphicscard. Its probably too low rated . Maybe swap it for something higher rated and see what happens
Ah, Voodoo ... loved those cards. They simply worked. They simply performed. They put out the best graphics. But then they got bored of being the best and started slacking hard.
I will have to re-watch this video again, maybe tomorrow when I am not so exhausted from work.
ATI Cards cannot cope witth high AGP Frequencys most of the time. With Radeon 9x00 Cards the limits seem to be about 80 - 90 MHz. (Blackscreen while Booting) Maybe on older Chips from ATI its worse.
Nvidia cards dont have this Problem at all (TNT2 to Gforce 4800). Reaching 100MHz with my geforce 4 on Nforce 2 pro was never a Problem. My Voodoo 3 3500 does not have Problems with high AGP clocks, too. Cannot speak for geforce 5 cards... Maybe the 440LX is bad, also. Had the ASUS P2L97, giving me headdaches every time
On ATI 9x00 cards, you do not gain any performance from overclocking the AGP Bus also. This maybe is happening on older chips, also. on Nvidia/ matrox and 3DFX Voodoo 3500 the performance scales very well.
that pci thing can be game port
21:36 у китайцев есть ATI Rage XL - это та же рейдж про. как раз на PCI.
Да, я её знаю, спасибо. У меня есть такая в AGP версии, но это не совсем тоже самое, что Rage Pro. Это немного доработаная версия с улучшенным техпроцессом и улучшенным функцоионалом. Я точно не знаю, как сильно они отличаются, но это была бы не плохая альтернатива. К сожалению пока такой нет, но в будущем надо будет попробовать.
@@necro_ware теперь мне стало интересно. этот комментарий вы написали через переводчик или вы знаете русский. 😅
@@digidoidit Да, я говорю на нескольких языках, включая Русский.
@@necro_ware прикольно. Eu tot comunic in 4 limbi. 😅 Ну как я понимаю русский английский немецкий. Ещё какие?
@@digidoidit Y español también, pero lamentablemente muy raramente.
Google says it is Parallel/Serial port....
Я бы вдобавок попробывал поставить win me и win 2000. Посмотреть, что бы они показали в плане нераспознаных устройств
How exactly do you disable SSE in software? o_O And what prevents any other software from detecting and using it again? AFAIK, SSE is completely broken in 440LX and is unfixable, it has something to do with the chipset being unable to transfer such long instructions between CPU/RAM/Cache.
Unfortunately, 440LX seems to be a very underwhelming chipset for any kind of P3 modification. You can run Tualatins on it, and if you're lucky, AGP will even work properly. But as soon as you run any SSE software, the system will crash. Win2000/XP won't even install. Late Win98 software will crash the system.
Slot-1 boards with 440LX, however, make perfect candidates for Pentium Pro builds with Socket8-to-Slot1 slockets with AGP port. You would need to manually mod the BIOS slightly.
Thank you! Yeah, this is my point I can deactivate SSE per software. This is documented and can be made by writing the right bits into CR4 control register, but as you said, nobody prevents other software from activating it again, that's why I don't think that it works. But my suspicion, that SSE could be the problem remains. I didn't find the way yet how to disable it permanently to see if this would change anything.
Ah, old Intel Chipsets and their pickiness with RAM...
You have to have them in a certain organisation to be detected correctly. If its the wrong one, you get half the Capacity, if it even works!
PRO TIP: use low capacity sticks to test Intel SDRAM Pentium2/3 Chipsets...
The i430VX for example only supports 16MB (maybe some 32MB?) Sticks for example...
OpenGL problem with the ATI...not every driver has or works with opengl. D3D should be fine!
Bonsoir 23:12 je pense que la carte mère n'as pas la bande passante suffisante pour faire marcher votre carte graphique qui sont généralement conçues pour des AGP x4 ou X8, vous pouvez essayer une *Nvidia Vanta* 16mo AGP X2 ou une *SiS* AGP X1 8mo qui est mieux tolérée par ce type de carte mère.
Anyone else wonder why the ATI Rage cards were so angry?
Because of them seeing Nvidia starting to dominate their market😁
2.nd hand air compressor? even better for the nature...
Using any SSE instructions will not work on 440LX and will hang on pc.
They called these GPUs RAGE for good reasons!!! Ugh!
What about Linux?
Yeah, ATI drivers of the time were horrible!
1:14 - конденсаторы тут не отсутствуют. Просто ушлые китайцы вместо двух или трёх твердотельных конденсаторов на меньшее напряжение и ёмкость ставят один с жидким электролитом для удешевления. Разработчики платы разводят плату по дизайн-гайдам: несколько твердотельных конденсаторов с низким суммарным ESR (поэтому мы видим три места для конденсаторов), дальше отдел "экономии" ставит один дешёвый конденсатор на эквивалентную ёмкость. Отсюда большой ESR, больше нагрев конденсатора и всё такое.
У меня были 2 одинаковые платы, на одной в каждой фазе стояли по 2 качественных твердотельных конденсатора на 2.5В 560мкФ, а в другой они были заменены на 6.3В 1000мкФ 105⁰C обычные электролитические для удешевления (точно на заводе, так как они были в менее ответственных местах на обеих платах возле PCI, USB...)
Moral of the story don't use Ati Rage cards! Just stick to 3dfx if you can or nvidia as a good alternative...
At 9:53 you claim that you think the fastest Coppermine Celeron is the 766. Did you mean to say the fastest at 66FSB? I know there are faster at 100FSB for sure, I've got a 1GHz Celly in my hand here!
Yeah, I meant 66FSB one indeed.
I think the fastest Coppermine was a Pentium III with 1.1 GHz and 100 MHz FSB if I am not mistaken.
@@kyoudaiken Yes at 100MHz FSB it was 1100MHz and at 133MHz FSB 1133MHz. Of course the 100MHz one is much more interesting, because it had higher multiplier and with a bit of luck could be overclocked to 133MHz FSB.
Maybe give Linux a try?
Chat GPT says:
The vendor ID 4348 corresponds to "XIO2000A/B PCI Express-to-PCI Bridge" by Texas Instruments. And the device ID 4447 typically indicates a specific model or variant of this bridge device.
Which once again proves that chat gpt doesn't know a thing. It just makes very good looking piles of BS.
@@aspinx Do you have evidence that it's wrong?
@@bstar777777 Isn't it obvious? What the heck PCI-Express to PCI bridge will do in the 90s hardware? There was no such thing as "PCI Express" at that time, it came years later.
@@aspinx Haha, you are correct, I overlooked that, I just saw PCI Bridge and assumed that was it.
Buy a ASUS P3B-F... :D
Intel: morrons since 1997.