Hi Adrian, when you looked the board up on the website, the image showed it with an AMI BIOS, maybe changing the BIOS might give you more control over the cache.
I remember the early days of SATA drives. Hybrid power was VERY common. Not many people had SATA power PSU's. so they did molex for a while to help bridge the gap. I remember getting all excited when I bought my first SATA only drive that no longer had molex.
Indeed. I had a local IBM office give me a bunch of SATA drives to test before they were commercially available to the public and they were Molex power and SATA interface only. I remember I had some issue and they had an IBM engineer phone me directly(!) and he was able to fix the issue in software. You won't get this kind of service today 😞
@@GeoffSeeley oh that's interesting. I've never seen any molex only. All the ones I saw were hybrid power. The only exclusives I've ever seen are SATA exclusives.
No more 4-pin Molex on DVD-ROM drives which are impossible to unplug. Thunderbolt is even better - all power, control,, and data on one small USC-C cable.
@@tenminutetokyo2643 I've never head of a thunderbolt DVD drive. But the interesting thing is there was a delay in getting SATA DVD drives. By the time they came out they were SATA power only.
As someone else mentioned, they made a 63Mhz PODP5. I had one and OCed it to 83 with no problem. Got it at a garage sale in the mid 90s for $1 - best upgrade of my life!
@Adrian: After recently restoring my own DX2-66 PC, I suspected your speed600 results were quite low, so I ran both it and cache check on my 486 to find out. My PC is a 5V AMD DX-2 66 with 256kb of L2 cache. The board is a PCChips M601 - also a VLB board. In cache check, my setup scores 16 microseconds from 1-8, and 26 microseconds from 16-256. In speed600, the CPU score is 317.11 Mhz and FPU is 488.00 Mhz. So from what I can tell, even your 486 DX-2 66 should be performing much better than it is.
I have a 486 motherboard with the Contaq 82C495A chipset (a Lion Computers Local ISA 486 that also has a "Made in USA" sticker on it), and I took a look at the Pentium Overdrive during a livestream. Both L1 and L2 caches were functioning with the Intel processors I tried (I went back and verified what cache check told me), however, my board only works in write through mode even though this is supposedly Contaq's write-back cache chipset. In fact, reviewing the spreadsheet of my benchmark results of several processors and L2 cache sizes reveal the following for Landmark Speed Test: 486DX2 66MHz, 64K L2 cache: CPU 276, FPU 476 Pentium Overdrive 83MHz, 64K or 256K L2 cache: CPU 508, FPU 1039 Am5x86 133MHz, 64K L2 cache: CPU 381, FPU 636 Am5x86 OC'ed to 160MHz, 64K L2 cache: CPU 501, FPU 840 I only tested the increased L2 cache on the POD processor. And yes, it appears as if my motherboard has an issue with the Am5x86 processor (which it wasn't designed to use but I am using it via one of those PNY upgrades that I got on eBay for relatively cheap). Looking at the CPU scores, it looks like the processor may be running at 3x multiplier even though I jumpered the upgrade for 4x... or maybe was the L1 cache disabled much like you saw with your POD??? Luckily I livestreamed the process and can go back and review the cache check results. Source (Time stamp is Landmark Speed Test with POD 83MHz with 64K L2 cache): ua-cam.com/users/liveIFo-8dhvSxc?feature=share&t=5063
Hello, I just noticed from an Adrian tweet about this issue, ge is getting less than 8MB/s on the main memory when testing with cachechk utility, how much are you getting?
You're right about the overdrive having some kind of protection if the fan isn't running. I had a similar situation happen to me with a Pentium overdrive, where processor was running super slow for no apparent reason. I pulled the cover off the computer and noticed that the fan was not turning. Once the fan was reconnected and spinning properly all was back to normal.
Pentium technically had separate Instruction and Data caches within it's L1 cache, rather than the Unified L1 that i486 ran... This might require a later BIOS or chipset that specifically recognises the difference. It also couldn't hurt to pull the L2 mobo cache chips temorarily just to test results on L1.
I'm betting this is more of a chipset issue. Chipset incompatibilities combined with price helped kill POD sales. One of AMDs 5x86 chips just made more sense as an upgrade. You could hunt down a regular AMI/Award Bios for the chipset (ideally one post POD release) and see if that helps.
The drive with both SATA and Molex power connectors were made in the transitional period between IDE drives and SATA drives. They were for people who upgraded their systems with SATA ports, but still had old PSUs with only Molex connectors. Also, do not connect both the SATA and Molex power connectors at the same time.
I miss the days when you could just pop in a new chip, and get 2x 3x or 4x (or more!) performance from the original chip, same motherboard. It seemed like magic at the time. The Socket 7 went even further. 75mhz all the way to 400mhz with MMX and 3DNow. That was an incredible time to live through when you'd get these huge performance gains just a couple years later with a CPU upgrade.
The Am4 socket is probably the closes we can get nowadays. Going from a R3 1200 to a R7 5800x3d is also a pretty big leap even if it's very unlikely that someone would go that path
The original Socket 7 was from 75-200 MHz, with some late board being able to go up to 233 MHz. Super Socket 7 extended that up to 570 MHz, but not all boards supported all speeds. CPU clock speeds were advancing so fast that chipsets and motherboard vendors couldn't keep up. Earlier Super 7 boards are less likely to support the later faster AMD K6/2 or 3 chips. There were of course the janky upgrade solutions by 3rd parties, but they were never guaranteed to work 100%. This video shows just one of the problems you can encounter, but the other is the FSB. An upgrade like the Evergreen Spectra that has a AMD K6/2 400 runs the FSB at 66 MHz, rather than the native 100 MHz of the CPU, slowing it down considerably. The whole Socket 7 / Super 7 era was a minefield of compatibility nightmares, I don't look back on it fondly.
@@harisalic2568 Just because both the R3 1200 and R7 5800X3D exist on the same socket, doesn't mean they will work on the same board. AM4 is a minefield of compatibility issues due to AMD promising too much from the AM4 platform. Socket longevity historically is not a good thing, and AM4 joins the ranks of Socket 7, Slot 1, PGA370 and LGA775 in terms of the number of compatibility headaches they all share. AMD had a lassiez-faire attitude when it came to motherboard firmware, and let vendors do whatever they wanted. This resulted in bloated graphically intense UEFI setup menus that used a considerable amount of ROM space on the limited 16M SPI ROM used on most motherboards. This came back to bite AMD in the ass by the time the Ryzen 3000 chips rolled around, their AGESA package that they supply to motherboard vendors that contains all of the microcode for their CPUs to work was getting too large to fit in those small 16M SPI ROMs with all of those fancy menus. In keeping with the lassiez-faire attitude they had, they still offered no real solutions and let board vendors do whatever they wanted to try and solve the problem. Board vendors then started either dropping support for older CPUs, or dropping support for random CPUs they deem not worthy of support. So this creates the problem for the buyer in that they can never know what AM4 motherboard they buy will support in terms of CPUs. So you can quite literally buy a brand new B550 or X570 board with an old firmware installed, that you have no control over, and it won't support the shiny new Ryzen 5000 series you bought. You'll either have to play the RMA lottery to get another board with hopefully a different firmware, or buy an older Ryzen CPU and hope that one is supported so you can flash your board to a more recent firmware release. I've literally had this happen four times with AM4 boards now, across four different chipsets. B350, x470, B550 and x570. In every case, the board shipped with a firmware that didn't support the CPU I was using, and I had to either RMA it, or find someone I could bum a CPU off of to get it reflashed. It doesn't help that the motherboard vendor claimed the CPUs I were trying to use were "officially supported", but not mentioning it required a specific firmware version.
@@GGigabiteM There's quite a few socket 7 motherboards that will fully support a K6-2 processor clocked at 400mhz. All it needs to support is 2.2V, and a 2x multiplier that the K6-2 interprets as 6x. I have one of them. Works great.
@@stevesether The 6x to 2x multiplier remapping was purely a thing done by AMD, it wasn't widely supported, and could be hazardous to your motherboard, potentially leading to its death. Getting a working multiplier does not equate to a motherboard supporting the CPU. If the BIOS doesn't have the correct microcode for the CPU, it may have erratic behavior, or not work at all. The cache not working properly was a big issue. Another big issue was power regulation. Boards without explicit support for K6/2 CPUs usually didn't have voltage selection that went down to 2.2v, and even if they did, power regulation for CPUs at that time was primitive. Many motherboards still used linear power regulators, which burn the difference between the input and output voltage as heat. The larger the distance between those two numbers, the more heat is produced. The boards with buck regulation also often were poorly configured, leading the mosfet to operate in the linear region, essentially the same as a linear regulator in regards to waste heat. Before slapping a K6/2 400 in an old Socket 7 board, these factors need to be researched thoroughly, else you may end up with a damaged motherboard.
Adrian, MR BIOS was usually an aftermarket bios people used for compatibility issues. I'd bet anything the MR BIOS on that board is not the appropriate build for that exact board, even if it's "close enough" for the chipset. As another commenter mentioned, using an official AMI BIOS might solve your cache issue. The Pentium Overdrive fan slowdown was actually a touted feature at the time - think about it though - lots of 486 boards that these were put in didn't even know the concept of a fan or cpu temperature, and the Pentium absolutely ran too hot to run without a working fan. Intel had to come up with some way to save the processor if the fan got dirty / stuck / died - since the motherboard and BIOS absolutely wouldn't do a thing about it. The funny thing is there's a lot of overdrives out there floating around without the fan on it - and replacement fans are hard to find. This was likely the opposite of Intel's original intent: save the expensive chip and just replace the cheap fan. Nowadays if you have a fanless PODP83, good luck getting it to run at 83 due to the fan slowdown routine and the proprietary heatsink.
Curious. What's the compatibility issue the MR BIOS was designed to solve? AMI was so ubiquitous back in that era, I'm surprised there were any issues at all.
"didn't even know the concept of a fan" Which is a interesting little detail. The fan is powered via the CPU pins since there is no header to plug any fan in. Explains what some of the additional pins are for.
A comment about the intel chip lifters. Don't pry the chip as you demonstrated. If you read the instructions in the intel guide they want you to push the lifter towards the chip. I worked in a computer shop and did quite a few 486 overdrive installations. Pulling the lifter can crack the edges of the socket. I saw it personally a few times. It is especially true 30 years later now that they are old and brittle. Push the lifter, don't pull it.
There could be a number of factors at play when benchmarking the performance. The motherboard BIOS has a problem correctly recognising the POD-83 (possibly around L1 cache) causing the benchmarking software to give false results, or the BIOS is correct and the software is at fault. I noticed Speed Mark 6.0 was copyrighted 1993, the POD-83 was released two years later. Perhaps it's mis-reading the CPU (note in the screenshot at 41:50, speed Mark claims the CPU is an 80386DX!).
That could make some sense if people like Phils or VSwitchZero haven't already tested the POD83 on Landmark v6 and get different and better results (his DX2 66 scores were already almost half of what other get) the BIOS is disabling the L1 on all the CPUs
That big sheet of foam may be handy for sorting all those loose ICs… now all we need to do is start scaring up a supply of IC tubes so they can be properly stored. The 4547 is a higher current version of the trusty 74LS47.
Pentium Overdrive CPU was a popular upgrade for the 486 computers especially to play Doom on it, as some games started to require a superscalar processor to run properly at decent framerate.
@@xsc1000 I agree, it just depends on how fast the 486 is running at - 66 to 133 MHz range for those ancient x86 processors were decent for this game as far as framerate go. At lower frequency, it's just a tad choppy. And Quake, yes, again, it also has to do with how fast the processors ran at, as GPUs weren't a thing just yet at that time, or if they were available, they were just very pricey (custom ASICs tend to be expensive).
The spikes on the inside of the telecom connector are used for clipping onto single conductor wire through the jacket. Might come in handy somewhere for checking. Voltages where the terminations are difficult to access.
After I figured out that all I had to do to overclock my Packard Bell Legend 660a (486SX-25) to 50 MHZ was a jumper setting, holy crap, did Sim City 2000 run amazingly! Was stable for YEARS. Don't know what happened to that beloved piece of crap computer. So many good memories.
This reminds me of the OptiFloppy and the Imation Superdisk being originally developed by Commodore for the Amiga. I can't remember who they partnered with but this also shares technology with the Iomega Zip disk. Of course by then they didn't need the optical track on the top of the disk. To answer your question though, yes that drive will in fact read and write regular 3-1/4" disks
I noticed that at 41:50 min of the video, during one of the tests with speed600, the background color turned blue, instead of red, and the detected CPU indicated an Intel 80386DX, instead of the 486 that was detected before. I don't know if this could indicate some problem or influence during the test or if I missed some detail. I just hope I could have helped if possible.
Just a reminder. Pentium's FPU is not much faster than 486's. The difference comes from the ability to run FPU asynchronously/separately from the rest of the CPU in Pentium. It's almost like you would have 2 cores in pentium.
I stil have the Galaxy S5 and original cable and agree that micro USB 3.0 port is annoying, but compatible with micro usb as well like most of the other phones in the galaxy line, just plug the micro USB cable into the wider portion of the jack on the phone/device.
There were 63 and 83 mhz Pentium overdrive processors. EDIT: The 486 overdrive processor marked 33 mhz indicates what the processor was intended to upgrade. This marking scheme didn't last long
Interesting, thanks. Suggestion, on your xt-ide flash card, create either dos menu or batch file(s) in root (or in path) to do the cd to change directories and type commands. Would save a lot of time for you and video as well as keystrokes.
We had a 486DX2 66 with 16 megs of RAM (in 1993!) and it was a beast... until Wing Commander III and IV. The Pentium Overdrive 83mhz upgrade really made a big difference.
Aha, yes, Wing Commander games were like the Crysis of their day. I remember the original Wing Commander was dog slow on anything but a beefy 386, which was too expensive for me at the time. As prices came down over the years, I was able to afford a second-hand 486 machine which finally allowed me to enjoy those games with a somewhat fluid framerate.
Dot's Pretzels are so good. The only flavors we have around here are regular, honey mustard, and cheese curls. I'm going to have to order some pork rinds. I bet those are amazing
It was a Samsung S5... One great thing though... It was water-resistant with a removable battery! Before iPhones were waterproof. I never dropped it in water, but I can see a big advantage if one does... Yank the battery out and dry it and no damage gets done...
And at 50:30 min, about the faulty connector pin, on the Floptical drive, it is not mentioned if a continuity test was done, so this could be a possible cause of the failure, if that pin is used for something.
13:37 (nice) They did also make 180mhz MMX overdrive chips as well, but the higher speed ones just upgraded a non-mmx pentium board to an mmx at higher clocks.. So for example, a Pentium-90mhz-to-150mhz would go up to an MMX-180mhz, or a Pentium-75 up to an MMX-150mhz.. I BELIEVE they had socket5 versions as well, though the frequency jump wasn't near as high, but it still did have multi-media-extensions (mmx) on a 486/socket5 system, something that could dramatically improve performance in certain workloads. Though in other workloads, like those heavily dependent on the bus for example, would remain more-or-less the same in terms of performance as the CPU was still very much limited by the bus as well as ram throughput While they're novel for sure, it's just as easy (and oftentimes cheaper) to build a socket7 retro system nowadays for we retro enthusiasts, and by avoiding using an overdrive, you're not bottlenecking yourself via the bus. There's no advantage a socket3/socket5 has over an early socket7 board as far as retro usage, except for maybe VLB, but there ARE socket7/pentium(non-mmx) boards that still have VLB slot(s), however those are exceedingly rare and their price reflects the scarcity. (That's why the two I have will be buried with me lol) I DO have a massive collection of other hardware (20+ socket3/5/7 motherboards, cards, sound, video, controllers, cpus, ram, HDDs, actual fully functional CRT monitors, that I've been trying to sell to a good home- I want to make sure the stuff goes to someone that's going to love it and not just flip it to earn a quick buck.
That EasyIDE USB to PATA (and SATA with adapters) is the BEST adapter. I fall back to it when even newer USB to SATA adapters don't work. It just WORKS.
Actually that connector is a very good thing - it resists the bending that causes desoldering from the board. And it is a very secure and tight fit for the cable.
Concerning Microid Research BIOS and shadowing: I guess there must be some ROM in UMA in order to be shadowed. If no ROM is present, the BIOS reports this area as "Vacant" and thus cannot be shadowed.
I've had one of those EasyIDE connectors since 2006. I still use it to this day. The USB SATA adaptor would be nice to have, it never exsisted back then though!
Yeah, that hard drive connector is weird. Seagate uses them on their USB 3 drives as well, and I actually had to go out and buy a standalone cable of that type when I got my Brother ADS-1200 desktop scanner, as it also uses this interface and did not come with a cable in the box.
The difference between the normal DX2/66 and the overdrive DX2/66 is the additional pin they have that is required in some motherboards with a soldered on CPU to disable the onboard CPU. There is two versions of them.
Well this is all fascinating. You should try all this on another period mobo, one that is a little more mainstream, say from a top-end IBM PS/Valuepoint 486. Those were good machines.
There are overdrive chips that go in the existing 486 CPU socket (ODPR) and overdrive chips intended for boards that had specific additional overdrive sockets (ODP). You read the part of the manual that refers to “Pentium overdrive processor socket” so I assume you have the latter, which should not be installed in the 486 CPU socket.
This difference was only an extra pin meant to disable a 486SX soldered to the board. They should both work fine in any motherboard without the soldered on CPU as far as I remember
We have Dots pretzels here in Illinois. Strangely enough, the almost 100% guaranteed location to get them is Ace Hardware stores. Some of the normal grocery stores have them too.
I believe the Pentium Overdrive had a 32K cache (which can operate in write-back mode on later motherboards, but only in the slower write-through mode on most 486 motherboards). The Pentium Overdrive should definitely be running faster than the 486DX2. (It's more similar to a 486DX4.) I know some early 486 boards aren't compatible with the Overdrive (because Intel changed the specs on the socket) but that doesn't explain why the cache on the 486 DX2 isn't working. I assume there must be something weird with that board or the BIOS.
The ones that didn't support the WB cache got nerfed worse than that: you didn't have any access to half the cache, so only the standard 16K (it had something to do with cache weirdness and how WT vs. WB cache worked).
I haven't managed to find a reasonably priced source for the anti-static foam. I bought some cheap parts bins that weren't anti-static, and I had intended on lining the drawers with some thin anti-static foam, but I shelved the project when I found out how much a big sheet of the stuff costs. I really should look around and see if I can find a cheaper sauce than the silicon-based electronics supplier.
The 'Telecom' clip has a 66-block jaw at the very front an a 'Bed-Of-Nails' connecter inside to clip onto telephone (thin insulation) wires without stripping (and not needing reinsulating afterwards) (I am a retired (2010 - 39yrs) telephone repairman/installer. I still have myy butt set, side tools and pole climbing gaffs)
Intel changed the pin out of the PODP *AFTER* they released the specification to vendors. Some of the first boards with the PODP processor socket don’t actually work correctly because the socket is “wired wrong” preventing proper cache operation. There was an interposer at the time to correct the issue, although usually the problem was that you couldn’t use the L2 cache, not the L1 cache. By the time VLB was around I would have thought any such issues would have been resolved. I’d try the official AMI bios and see if that works any better.
0:40 YEAH! I'd get behind adding a map into these mail call videos! That sounds cool! And if you digitalize it, then you could simply update it from software.
The IBM "Deathstar" moniker was specifically for the 75GXP models, launched in 2001 and available in sizes from 15GB to 75GB and leading to an US class-action lawsuits and a settlement in 2005 not admitting any fault but supposedely paying out $100 to any US user with a failed 75GXP. In my experience back in the days earlier deskstar models tends to be some of the more reliable drives from that era and IIRC the 75GXP was the fastest and IIRC largest disk you could buy when it was launched so they sold really well for IBM. Documents from the lawsuit proves IBM knew they were failing at 10x the rate it should and was trying to evade responsibility and just keep selling them! They really did deserve that moniker. IBM really did a number on their own business with that one, they sold their entire drive business to Hitachi two years later due to the fallout.
I had 2 1G ones, and yes both did eventually fail, one quietly, and the other was just taking longer and longer to read data, till the Adaptec SCSI controller eventually timed out with a BSOD. Had a similar one with a Seagate drive, cured by putting in the freezer for 3 hours in a plastic bag, and making a copy fast before it warmed up. Replaced under warranty by the local agents, though to fail it in the walk in freezer they used as office ( receptionist was wearing a jersey, when it was 30C outside) I had to put the top back on that plastic Seashield case to warm the drive up to 30C, when it would fail.
I still remember when I got my 14GXP. First IDE 7200rpm drive, I think with ATA-33 too. SO FAST for an IDE drive. No click-o-death on that one. I've still got it around here somewhere. :D
Yes, the DeskStar drives were very good excluding that specific window of 75GXP model units; I suppose that makes them a great find for us now as most dum dums avoid the 'IBM' and 'DeskStar' HDD branded drives entirely because of a stupid internet meme.
IKEA CR2032 battery? LOL! That's my weird takeaway from this video. Floptical...WOW. I never knew that was a thing. And the precursor to the Iomega Zip dive. No kidding. Mind BLOWN. Thanks for another great video. Interested to see if you find another 486 mobo to test everything. I've got one but it's a spare for an old voicemail PC. Still working in 2023! :D
Even if the upgrade CPU is a DX instead of a specific 487 or overdrive you can get the system to disable the onboard SX CPU by grounding that extra pin (or is it 5v, better you double check hehe but it works, I will place bets that you can put a normal 486DX without the additional pin on any SX system by just wiring that pin correctly to signal the onboard CPU to go off)
Shadow and cash are different. Cash is a dynamic thing, what gets stored is cash depends on what the CPU accesses. Where as a "shadow" of a ROM includes ALL the ROM even the parts that may not be accessed. Also with a cash it only increases access to the second time some data is used, the first time is alway ways taken from main memory.
Shadow RAM on a lot of video cards did not work because the card used a memory mapped section next to the ROM as video memory, and caching the area would result in the card not being able to display video, as it would not be writing to the right area in the card memory. Common on cards that had extended modes, which did not fit the standard memory mapping. Not all overdrive processors worked well, especially on all motherboards. On board cache on the die was always preferred, and the overdrive processors would often disable the cache on power up, as it was doing cache control, and would only use main memory.
There are pentium overdrives that run at 63MHz as well, but the 83MHz ones are more common. Those run at 2.5x on a 25MHz motherboard (the 83 runs at 2.5X on 33MHz bus). It's interesting that you can run these overdrives at faster speeds than their rated one as well; for instance I am running one of them at 100MHz on a 486 motherboard with a 40MHz FSB. They usually need a little voltage boost to run at the higher speeds, but they are fairly tolerant of overvoltages. To boost the core voltage you have to modify the on-board voltage regulator but that is not too hard to do, all you need to do is insert one (or more) diodes in series with the reference terminal of the VRM chip, each diode adds an extra 0.7V to the regulated voltage. I run mine with one diode which gives a core voltage of 3.3+0.7=4V, with no issues. It doesn't overheat and runs perfectly stable. I know people that run them even at 115-120MHz or more.
The various 486's had slightly different pinouts, some of which have to do with on-chip cache functionality. You'll have to check that your motherboard is set up correctly for the chip you're using.
It was the Samsung Galaxy S5 and Note 3 that have that micro usb 3.0 I’ve owned both phones at one point. Oddly enough most people thought you needed a 3.0 cable to charge the phone so the 3.0 connector was dropped from the next generation. I’ll never understand why people assume you can’t use a 2.0 cable on a 3.0 connector. Literally remember a customer inside Walmart asking for a 3.0 micro usb cable to charge a Samsung phone. At the time not only was that type of cable hard to find but but was generally double or even triple the price. Needless to say that customer went away angry as they couldn’t buy said cable because the only one they sold was out of stock. Oddly enough no amount talking to them was able to convince them that they could buy and use a standard micro usb to charge said phone
I had this CPU and a similar motherboard, I don't remember the exact board, its been years, but I accidentally ran across a jumper on the motherboard that set the CPU multiplier, I removed the jumper, and the System Speed Doubled. I believe the bus clock divisor has to be disabled.
Hi Adrian, the Pentium 1 have L1 cache have 16 KB (internal CPU), no L2 cache, and L3 cache have 128 KB found on your motherboard. The 80486 family CPU have L1 cache have 8 KB memory but few 80486 have 16 KB memory.
It was the Note5, I loved that thing, I Lost the Stylus while writing a Recipie down while working as a chef lol Nobody ever told you this, but you could use a standard Micro-USB Cord!!! It will still work 100% compatible with USB 2: I learned there somewhere around 2015. Blew My Mind that it wasn't common Knowledge. I had no idea lol
Those SK series chip appear to be equivalents to 74 series TTL chips, for example a 7447 is a BCD to 7 segment decoder, just like that 4547. Actually, I am wondering if the 4xxx means it's CMOS, so equivalent to a 74C47
Problem with L1 cache is definitelly somewhere on board. Inside 486 CPU you can only control cache disable (CD) bit and not write-trough (NW) bit in CR0 register. This can be probably enabled/checked by debug.exe. Rest of L1 cache control is in chipset. Also BIOS usually uses L1 cache as RAM before DRAM is detected and configured. It seems the chipset can be either 82C596A or 82C597 and it seems to be cypress (cypress later bought contaq). There seems to be a datasheet for CY82C597 with registers description how to control cache (also L1 write back configuration).
I have an external Maxell LS-120 for parallel port and one LS-120 disk. Years ago, i was able to make it work under Windows 98. Due to the parallel port connection, it is pretty slow. But what's interesting, it is able to format a regular 1.44MByte floppy to store ~32MByte. Cool device, but not very usable. These LS-120 floppies also seem to be very unobtanium. Another interesting aspect about that floppy disk is, you can see little squares on it. Not sure if that information is useful, but that's what i figured out, when i played with it.
Hi Adrian, when you looked the board up on the website, the image showed it with an AMI BIOS, maybe changing the BIOS might give you more control over the cache.
I was gonna say the same thing. AMI BIOS vs Mr BIOS.
I remember the early days of SATA drives. Hybrid power was VERY common. Not many people had SATA power PSU's. so they did molex for a while to help bridge the gap. I remember getting all excited when I bought my first SATA only drive that no longer had molex.
Indeed. I had a local IBM office give me a bunch of SATA drives to test before they were commercially available to the public and they were Molex power and SATA interface only. I remember I had some issue and they had an IBM engineer phone me directly(!) and he was able to fix the issue in software. You won't get this kind of service today 😞
@@GeoffSeeley oh that's interesting. I've never seen any molex only. All the ones I saw were hybrid power. The only exclusives I've ever seen are SATA exclusives.
No more 4-pin Molex on DVD-ROM drives which are impossible to unplug. Thunderbolt is even better - all power, control,, and data on one small USC-C cable.
@@tenminutetokyo2643 I've never head of a thunderbolt DVD drive. But the interesting thing is there was a delay in getting SATA DVD drives. By the time they came out they were SATA power only.
There were also Western digital SATA hard drives with dual power inputs.
As someone else mentioned, they made a 63Mhz PODP5. I had one and OCed it to 83 with no problem. Got it at a garage sale in the mid 90s for $1 - best upgrade of my life!
@Adrian: After recently restoring my own DX2-66 PC, I suspected your speed600 results were quite low, so I ran both it and cache check on my 486 to find out. My PC is a 5V AMD DX-2 66 with 256kb of L2 cache. The board is a PCChips M601 - also a VLB board. In cache check, my setup scores 16 microseconds from 1-8, and 26 microseconds from 16-256. In speed600, the CPU score is 317.11 Mhz and FPU is 488.00 Mhz. So from what I can tell, even your 486 DX-2 66 should be performing much better than it is.
I remember pc chips. Weren't they the company that did the fake cache chips?
There are many more variables. Chip sets and other things play a big part.
I have a 486 motherboard with the Contaq 82C495A chipset (a Lion Computers Local ISA 486 that also has a "Made in USA" sticker on it), and I took a look at the Pentium Overdrive during a livestream. Both L1 and L2 caches were functioning with the Intel processors I tried (I went back and verified what cache check told me), however, my board only works in write through mode even though this is supposedly Contaq's write-back cache chipset.
In fact, reviewing the spreadsheet of my benchmark results of several processors and L2 cache sizes reveal the following for Landmark Speed Test:
486DX2 66MHz, 64K L2 cache: CPU 276, FPU 476
Pentium Overdrive 83MHz, 64K or 256K L2 cache: CPU 508, FPU 1039
Am5x86 133MHz, 64K L2 cache: CPU 381, FPU 636
Am5x86 OC'ed to 160MHz, 64K L2 cache: CPU 501, FPU 840
I only tested the increased L2 cache on the POD processor.
And yes, it appears as if my motherboard has an issue with the Am5x86 processor (which it wasn't designed to use but I am using it via one of those PNY upgrades that I got on eBay for relatively cheap). Looking at the CPU scores, it looks like the processor may be running at 3x multiplier even though I jumpered the upgrade for 4x... or maybe was the L1 cache disabled much like you saw with your POD??? Luckily I livestreamed the process and can go back and review the cache check results.
Source (Time stamp is Landmark Speed Test with POD 83MHz with 64K L2 cache): ua-cam.com/users/liveIFo-8dhvSxc?feature=share&t=5063
Hello, I just noticed from an Adrian tweet about this issue, ge is getting less than 8MB/s on the main memory when testing with cachechk utility, how much are you getting?
I have checked with Landmark System Speed 6.00 (speed600) and my DX2-66 result is:
CPU: 316 MHz
FPU: 453 MHZ
YAWN YAWN...
@@MothKeeperwtf is your problem?
🤣🤣🤣 toilet.@@AureliusR
You're right about the overdrive having some kind of protection if the fan isn't running. I had a similar situation happen to me with a Pentium overdrive, where processor was running super slow for no apparent reason. I pulled the cover off the computer and noticed that the fan was not turning. Once the fan was reconnected and spinning properly all was back to normal.
Pentium technically had separate Instruction and Data caches within it's L1 cache, rather than the Unified L1 that i486 ran... This might require a later BIOS or chipset that specifically recognises the difference.
It also couldn't hurt to pull the L2 mobo cache chips temorarily just to test results on L1.
I'm betting this is more of a chipset issue. Chipset incompatibilities combined with price helped kill POD sales. One of AMDs 5x86 chips just made more sense as an upgrade. You could hunt down a regular AMI/Award Bios for the chipset (ideally one post POD release) and see if that helps.
The drive with both SATA and Molex power connectors were made in the transitional period between IDE drives and SATA drives. They were for people who upgraded their systems with SATA ports, but still had old PSUs with only Molex connectors. Also, do not connect both the SATA and Molex power connectors at the same time.
I miss the days when you could just pop in a new chip, and get 2x 3x or 4x (or more!) performance from the original chip, same motherboard. It seemed like magic at the time.
The Socket 7 went even further. 75mhz all the way to 400mhz with MMX and 3DNow. That was an incredible time to live through when you'd get these huge performance gains just a couple years later with a CPU upgrade.
The Am4 socket is probably the closes we can get nowadays. Going from a R3 1200 to a R7 5800x3d is also a pretty big leap even if it's very unlikely that someone would go that path
The original Socket 7 was from 75-200 MHz, with some late board being able to go up to 233 MHz. Super Socket 7 extended that up to 570 MHz, but not all boards supported all speeds. CPU clock speeds were advancing so fast that chipsets and motherboard vendors couldn't keep up. Earlier Super 7 boards are less likely to support the later faster AMD K6/2 or 3 chips.
There were of course the janky upgrade solutions by 3rd parties, but they were never guaranteed to work 100%. This video shows just one of the problems you can encounter, but the other is the FSB. An upgrade like the Evergreen Spectra that has a AMD K6/2 400 runs the FSB at 66 MHz, rather than the native 100 MHz of the CPU, slowing it down considerably.
The whole Socket 7 / Super 7 era was a minefield of compatibility nightmares, I don't look back on it fondly.
@@harisalic2568 Just because both the R3 1200 and R7 5800X3D exist on the same socket, doesn't mean they will work on the same board. AM4 is a minefield of compatibility issues due to AMD promising too much from the AM4 platform. Socket longevity historically is not a good thing, and AM4 joins the ranks of Socket 7, Slot 1, PGA370 and LGA775 in terms of the number of compatibility headaches they all share.
AMD had a lassiez-faire attitude when it came to motherboard firmware, and let vendors do whatever they wanted. This resulted in bloated graphically intense UEFI setup menus that used a considerable amount of ROM space on the limited 16M SPI ROM used on most motherboards. This came back to bite AMD in the ass by the time the Ryzen 3000 chips rolled around, their AGESA package that they supply to motherboard vendors that contains all of the microcode for their CPUs to work was getting too large to fit in those small 16M SPI ROMs with all of those fancy menus.
In keeping with the lassiez-faire attitude they had, they still offered no real solutions and let board vendors do whatever they wanted to try and solve the problem. Board vendors then started either dropping support for older CPUs, or dropping support for random CPUs they deem not worthy of support. So this creates the problem for the buyer in that they can never know what AM4 motherboard they buy will support in terms of CPUs.
So you can quite literally buy a brand new B550 or X570 board with an old firmware installed, that you have no control over, and it won't support the shiny new Ryzen 5000 series you bought. You'll either have to play the RMA lottery to get another board with hopefully a different firmware, or buy an older Ryzen CPU and hope that one is supported so you can flash your board to a more recent firmware release.
I've literally had this happen four times with AM4 boards now, across four different chipsets. B350, x470, B550 and x570. In every case, the board shipped with a firmware that didn't support the CPU I was using, and I had to either RMA it, or find someone I could bum a CPU off of to get it reflashed. It doesn't help that the motherboard vendor claimed the CPUs I were trying to use were "officially supported", but not mentioning it required a specific firmware version.
@@GGigabiteM There's quite a few socket 7 motherboards that will fully support a K6-2 processor clocked at 400mhz. All it needs to support is 2.2V, and a 2x multiplier that the K6-2 interprets as 6x.
I have one of them. Works great.
@@stevesether The 6x to 2x multiplier remapping was purely a thing done by AMD, it wasn't widely supported, and could be hazardous to your motherboard, potentially leading to its death.
Getting a working multiplier does not equate to a motherboard supporting the CPU. If the BIOS doesn't have the correct microcode for the CPU, it may have erratic behavior, or not work at all. The cache not working properly was a big issue.
Another big issue was power regulation. Boards without explicit support for K6/2 CPUs usually didn't have voltage selection that went down to 2.2v, and even if they did, power regulation for CPUs at that time was primitive.
Many motherboards still used linear power regulators, which burn the difference between the input and output voltage as heat. The larger the distance between those two numbers, the more heat is produced. The boards with buck regulation also often were poorly configured, leading the mosfet to operate in the linear region, essentially the same as a linear regulator in regards to waste heat.
Before slapping a K6/2 400 in an old Socket 7 board, these factors need to be researched thoroughly, else you may end up with a damaged motherboard.
You guys remember the over drive ready boards with the extra Big Blue Socket with white letters on it? :)
Adrian, MR BIOS was usually an aftermarket bios people used for compatibility issues. I'd bet anything the MR BIOS on that board is not the appropriate build for that exact board, even if it's "close enough" for the chipset. As another commenter mentioned, using an official AMI BIOS might solve your cache issue.
The Pentium Overdrive fan slowdown was actually a touted feature at the time - think about it though - lots of 486 boards that these were put in didn't even know the concept of a fan or cpu temperature, and the Pentium absolutely ran too hot to run without a working fan. Intel had to come up with some way to save the processor if the fan got dirty / stuck / died - since the motherboard and BIOS absolutely wouldn't do a thing about it.
The funny thing is there's a lot of overdrives out there floating around without the fan on it - and replacement fans are hard to find. This was likely the opposite of Intel's original intent: save the expensive chip and just replace the cheap fan. Nowadays if you have a fanless PODP83, good luck getting it to run at 83 due to the fan slowdown routine and the proprietary heatsink.
I agree, first thing I'd try would be AMI BIOS
You can run any cheap NB/SB heat sink on them with a thermal adhesive strip.
Curious. What's the compatibility issue the MR BIOS was designed to solve? AMI was so ubiquitous back in that era, I'm surprised there were any issues at all.
"didn't even know the concept of a fan" Which is a interesting little detail. The fan is powered via the CPU pins since there is no header to plug any fan in. Explains what some of the additional pins are for.
@@agurdel Yep, I had a DX2/66 at one stage which had a small fan and it was powered from a Molex connection on the PSU.
A comment about the intel chip lifters. Don't pry the chip as you demonstrated. If you read the instructions in the intel guide they want you to push the lifter towards the chip. I worked in a computer shop and did quite a few 486 overdrive installations. Pulling the lifter can crack the edges of the socket. I saw it personally a few times. It is especially true 30 years later now that they are old and brittle. Push the lifter, don't pull it.
There could be a number of factors at play when benchmarking the performance. The motherboard BIOS has a problem correctly recognising the POD-83 (possibly around L1 cache) causing the benchmarking software to give false results, or the BIOS is correct and the software is at fault. I noticed Speed Mark 6.0 was copyrighted 1993, the POD-83 was released two years later. Perhaps it's mis-reading the CPU (note in the screenshot at 41:50, speed Mark claims the CPU is an 80386DX!).
That could make some sense if people like Phils or VSwitchZero haven't already tested the POD83 on Landmark v6 and get different and better results (his DX2 66 scores were already almost half of what other get) the BIOS is disabling the L1 on all the CPUs
That big sheet of foam may be handy for sorting all those loose ICs… now all we need to do is start scaring up a supply of IC tubes so they can be properly stored. The 4547 is a higher current version of the trusty 74LS47.
I love your channel, I have never been into computer repair or building but these are very interesting, I could watch all day.
Pentium Overdrive CPU was a popular upgrade for the 486 computers especially to play Doom on it, as some games started to require a superscalar processor to run properly at decent framerate.
Doom was OK on 486. Did you mean Quake?
@@xsc1000 I agree, it just depends on how fast the 486 is running at - 66 to 133 MHz range for those ancient x86 processors were decent for this game as far as framerate go. At lower frequency, it's just a tad choppy. And Quake, yes, again, it also has to do with how fast the processors ran at, as GPUs weren't a thing just yet at that time, or if they were available, they were just very pricey (custom ASICs tend to be expensive).
The spikes on the inside of the telecom connector are used for clipping onto single conductor wire through the jacket. Might come in handy somewhere for checking. Voltages where the terminations are difficult to access.
After I figured out that all I had to do to overclock my Packard Bell Legend 660a (486SX-25) to 50 MHZ was a jumper setting, holy crap, did Sim City 2000 run amazingly! Was stable for YEARS. Don't know what happened to that beloved piece of crap computer. So many good memories.
This reminds me of the OptiFloppy and the Imation Superdisk being originally developed by Commodore for the Amiga. I can't remember who they partnered with but this also shares technology with the Iomega Zip disk. Of course by then they didn't need the optical track on the top of the disk. To answer your question though, yes that drive will in fact read and write regular 3-1/4" disks
I noticed that at 41:50 min of the video, during one of the tests with speed600, the background color turned blue, instead of red, and the detected CPU indicated an Intel 80386DX, instead of the 486 that was detected before. I don't know if this could indicate some problem or influence during the test or if I missed some detail. I just hope I could have helped if possible.
Just a reminder. Pentium's FPU is not much faster than 486's. The difference comes from the ability to run FPU asynchronously/separately from the rest of the CPU in Pentium. It's almost like you would have 2 cores in pentium.
To make sure your boot floppies are always accessible, use a permanent magnet to attack them to a file cabinet. Easy peasy!
^-^
No!
I stil have the Galaxy S5 and original cable and agree that micro USB 3.0 port is annoying, but compatible with micro usb as well like most of the other phones in the galaxy line, just plug the micro USB cable into the wider portion of the jack on the phone/device.
There were 63 and 83 mhz Pentium overdrive processors. EDIT: The 486 overdrive processor marked 33 mhz indicates what the processor was intended to upgrade. This marking scheme didn't last long
Yeah, both had a hardwired 2.5 multi, but the first was intended as a replacement for a 25 mhz 486, the other was for a 33mhz 486.
That explains it then. I didn't know that.
@@Retrohertz I bet the 63 mhz versions will work fine in a 33mhz board btw.
@@ironhead2008 Can confirm that they do.
@@dennisp.2147 Figures, I doubt they would bother binning them that tightly, they were probably relying on the average end user not knowing better.
Interesting, thanks. Suggestion, on your xt-ide flash card, create either dos menu or batch file(s) in root (or in path) to do the cd to change directories and type commands. Would save a lot of time for you and video as well as keystrokes.
Adrian - you were right about the Dot's pretzels - picked up a pack of the honey mustard and they are great!! :)
3:20 can confirm, I’m also an Oregonian and those pretzels are THE BEST.
Just about all early SATA drives had 12V molex for power because not everyone switched over yet. Needed both for the transition period.
The POD83 is for Socket 3, while the DX2 and DX4 was for Socket 2. That's why you are seeing additional pins on the POD83.
We had a 486DX2 66 with 16 megs of RAM (in 1993!) and it was a beast... until Wing Commander III and IV. The Pentium Overdrive 83mhz upgrade really made a big difference.
Aha, yes, Wing Commander games were like the Crysis of their day. I remember the original Wing Commander was dog slow on anything but a beefy 386, which was too expensive for me at the time. As prices came down over the years, I was able to afford a second-hand 486 machine which finally allowed me to enjoy those games with a somewhat fluid framerate.
That pushed in pin means that the SCSI cable was connected backwards... I'd wager the card and/or drive had something fry because of that...
You did one change that could have an effect on the cache situation. You disabled Maximize Ext-Mem (REMAP).
🤓
Dot's Pretzels are so good. The only flavors we have around here are regular, honey mustard, and cheese curls. I'm going to have to order some pork rinds. I bet those are amazing
It was a Samsung S5... One great thing though... It was water-resistant with a removable battery! Before iPhones were waterproof. I never dropped it in water, but I can see a big advantage if one does... Yank the battery out and dry it and no damage gets done...
And at 50:30 min, about the faulty connector pin, on the Floptical drive, it is not mentioned if a continuity test was done, so this could be a possible cause of the failure, if that pin is used for something.
I can get Dot's pretzels where I live in Western NY. They were a big hit at a friend's party, everyone loved them! 😀
It was the Note 3... And I LOVED it!!!
13:37 (nice) They did also make 180mhz MMX overdrive chips as well, but the higher speed ones just upgraded a non-mmx pentium board to an mmx at higher clocks.. So for example, a Pentium-90mhz-to-150mhz would go up to an MMX-180mhz, or a Pentium-75 up to an MMX-150mhz.. I BELIEVE they had socket5 versions as well, though the frequency jump wasn't near as high, but it still did have multi-media-extensions (mmx) on a 486/socket5 system, something that could dramatically improve performance in certain workloads. Though in other workloads, like those heavily dependent on the bus for example, would remain more-or-less the same in terms of performance as the CPU was still very much limited by the bus as well as ram throughput
While they're novel for sure, it's just as easy (and oftentimes cheaper) to build a socket7 retro system nowadays for we retro enthusiasts, and by avoiding using an overdrive, you're not bottlenecking yourself via the bus. There's no advantage a socket3/socket5 has over an early socket7 board as far as retro usage, except for maybe VLB, but there ARE socket7/pentium(non-mmx) boards that still have VLB slot(s), however those are exceedingly rare and their price reflects the scarcity. (That's why the two I have will be buried with me lol)
I DO have a massive collection of other hardware (20+ socket3/5/7 motherboards, cards, sound, video, controllers, cpus, ram, HDDs, actual fully functional CRT monitors, that I've been trying to sell to a good home- I want to make sure the stuff goes to someone that's going to love it and not just flip it to earn a quick buck.
46:45 I note that the board in the photo has AMIBIOS.
0:28 Friends asked back in May on my Discord server..."AJ WHAT have you done to Adrian??!!!!" Ah, well...whatever it is, I love it!
That EasyIDE USB to PATA (and SATA with adapters) is the BEST adapter. I fall back to it when even newer USB to SATA adapters don't work. It just WORKS.
Actually that connector is a very good thing - it resists the bending that causes desoldering from the board. And it is a very secure and tight fit for the cable.
Concerning Microid Research BIOS and shadowing: I guess there must be some ROM in UMA in order to be shadowed. If no ROM is present, the BIOS reports this area as "Vacant" and thus cannot be shadowed.
USB3 -> Sata usually only ever worked for SSD's for me
Also, funny enough... we have a production machine still running a 386.
Honey Mustard Dot's pretzels are the best IMO. Try to always keep a bag around. They sell them here in Chicagoland, Illinois.
I've had one of those EasyIDE connectors since 2006. I still use it to this day. The USB SATA adaptor would be nice to have, it never exsisted back then though!
First of all you should use AMI bios, like it was originally on this board. Also check switches of L2 cache, may be its set incorrectly.
Yeah, that hard drive connector is weird. Seagate uses them on their USB 3 drives as well, and I actually had to go out and buy a standalone cable of that type when I got my Brother ADS-1200 desktop scanner, as it also uses this interface and did not come with a cable in the box.
I see I’m not the only one who likes those pretzels! 😊
doublecheck JP3 and JP4 again - this are different for P24T...
You are right up4 must be close on P24
@@christophejournoud2773 jp3 is also different
THIS! Jumpers need set to P24T
I would definitely try the P24T setting with the jumpers!
P24c is 486dx4 75 or 100mhz processor, and p24t is Pentium overdrive processor.
I had one in my SGI Indy quite a long time ago now. I just used it as a normal floppy drive.
We have Dot's Pretzels here in Kentucky. They have a limited edition Cinnamon Sugar Pretzel currently that is incredible.
The difference between the normal DX2/66 and the overdrive DX2/66 is the additional pin they have that is required in some motherboards with a soldered on CPU to disable the onboard CPU.
There is two versions of them.
Well this is all fascinating. You should try all this on another period mobo, one that is a little more mainstream, say from a top-end IBM PS/Valuepoint 486. Those were good machines.
There are overdrive chips that go in the existing 486 CPU socket (ODPR) and overdrive chips intended for boards that had specific additional overdrive sockets (ODP).
You read the part of the manual that refers to “Pentium overdrive processor socket” so I assume you have the latter, which should not be installed in the 486 CPU socket.
This difference was only an extra pin meant to disable a 486SX soldered to the board. They should both work fine in any motherboard without the soldered on CPU as far as I remember
We have Dots pretzels here in Illinois. Strangely enough, the almost 100% guaranteed location to get them is Ace Hardware stores. Some of the normal grocery stores have them too.
I believe the Pentium Overdrive had a 32K cache (which can operate in write-back mode on later motherboards, but only in the slower write-through mode on most 486 motherboards). The Pentium Overdrive should definitely be running faster than the 486DX2. (It's more similar to a 486DX4.)
I know some early 486 boards aren't compatible with the Overdrive (because Intel changed the specs on the socket) but that doesn't explain why the cache on the 486 DX2 isn't working. I assume there must be something weird with that board or the BIOS.
The ones that didn't support the WB cache got nerfed worse than that: you didn't have any access to half the cache, so only the standard 16K (it had something to do with cache weirdness and how WT vs. WB cache worked).
I haven't managed to find a reasonably priced source for the anti-static foam. I bought some cheap parts bins that weren't anti-static, and I had intended on lining the drawers with some thin anti-static foam, but I shelved the project when I found out how much a big sheet of the stuff costs. I really should look around and see if I can find a cheaper sauce than the silicon-based electronics supplier.
Word fun - "floptical" is an example of what is called a portmanteau.
The 'Telecom' clip has a 66-block jaw at the very front an a 'Bed-Of-Nails' connecter inside to clip onto telephone (thin insulation) wires without stripping (and not needing reinsulating afterwards)
(I am a retired (2010 - 39yrs) telephone repairman/installer. I still have myy butt set, side tools and pole climbing gaffs)
I only discovered Dots Pretzels recently in Wisconsin, but they are thoroughly addictive!
Intel changed the pin out of the PODP *AFTER* they released the specification to vendors. Some of the first boards with the PODP processor socket don’t actually work correctly because the socket is “wired wrong” preventing proper cache operation. There was an interposer at the time to correct the issue, although usually the problem was that you couldn’t use the L2 cache, not the L1 cache.
By the time VLB was around I would have thought any such issues would have been resolved. I’d try the official AMI bios and see if that works any better.
0:40 YEAH! I'd get behind adding a map into these mail call videos! That sounds cool! And if you digitalize it, then you could simply update it from software.
The IBM "Deathstar" moniker was specifically for the 75GXP models, launched in 2001 and available in sizes from 15GB to 75GB and leading to an US class-action lawsuits and a settlement in 2005 not admitting any fault but supposedely paying out $100 to any US user with a failed 75GXP. In my experience back in the days earlier deskstar models tends to be some of the more reliable drives from that era and IIRC the 75GXP was the fastest and IIRC largest disk you could buy when it was launched so they sold really well for IBM. Documents from the lawsuit proves IBM knew they were failing at 10x the rate it should and was trying to evade responsibility and just keep selling them! They really did deserve that moniker.
IBM really did a number on their own business with that one, they sold their entire drive business to Hitachi two years later due to the fallout.
I had 2 1G ones, and yes both did eventually fail, one quietly, and the other was just taking longer and longer to read data, till the Adaptec SCSI controller eventually timed out with a BSOD. Had a similar one with a Seagate drive, cured by putting in the freezer for 3 hours in a plastic bag, and making a copy fast before it warmed up. Replaced under warranty by the local agents, though to fail it in the walk in freezer they used as office ( receptionist was wearing a jersey, when it was 30C outside) I had to put the top back on that plastic Seashield case to warm the drive up to 30C, when it would fail.
I still remember when I got my 14GXP. First IDE 7200rpm drive, I think with ATA-33 too. SO FAST for an IDE drive. No click-o-death on that one. I've still got it around here somewhere. :D
Yes, the DeskStar drives were very good excluding that specific window of 75GXP model units; I suppose that makes them a great find for us now as most dum dums avoid the 'IBM' and 'DeskStar' HDD branded drives entirely because of a stupid internet meme.
Galaxy S5 used that usb. I loved the connector.
IKEA CR2032 battery? LOL! That's my weird takeaway from this video.
Floptical...WOW. I never knew that was a thing. And the precursor to the Iomega Zip dive. No kidding. Mind BLOWN.
Thanks for another great video. Interested to see if you find another 486 mobo to test everything. I've got one but it's a spare for an old voicemail PC. Still working in 2023! :D
Even if the upgrade CPU is a DX instead of a specific 487 or overdrive you can get the system to disable the onboard SX CPU by grounding that extra pin (or is it 5v, better you double check hehe but it works, I will place bets that you can put a normal 486DX without the additional pin on any SX system by just wiring that pin correctly to signal the onboard CPU to go off)
Shadow and cash are different. Cash is a dynamic thing, what gets stored is cash depends on what the CPU accesses. Where as a "shadow" of a ROM includes ALL the ROM even the parts that may not be accessed. Also with a cash it only increases access to the second time some data is used, the first time is alway ways taken from main memory.
Shadow RAM on a lot of video cards did not work because the card used a memory mapped section next to the ROM as video memory, and caching the area would result in the card not being able to display video, as it would not be writing to the right area in the card memory. Common on cards that had extended modes, which did not fit the standard memory mapping.
Not all overdrive processors worked well, especially on all motherboards. On board cache on the die was always preferred, and the overdrive processors would often disable the cache on power up, as it was doing cache control, and would only use main memory.
There are pentium overdrives that run at 63MHz as well, but the 83MHz ones are more common. Those run at 2.5x on a 25MHz motherboard (the 83 runs at 2.5X on 33MHz bus). It's interesting that you can run these overdrives at faster speeds than their rated one as well; for instance I am running one of them at 100MHz on a 486 motherboard with a 40MHz FSB. They usually need a little voltage boost to run at the higher speeds, but they are fairly tolerant of overvoltages. To boost the core voltage you have to modify the on-board voltage regulator but that is not too hard to do, all you need to do is insert one (or more) diodes in series with the reference terminal of the VRM chip, each diode adds an extra 0.7V to the regulated voltage. I run mine with one diode which gives a core voltage of 3.3+0.7=4V, with no issues. It doesn't overheat and runs perfectly stable. I know people that run them even at 115-120MHz or more.
You may find that the locations the BIOS can't "Shadow" are locations that do not have ROM or that the ROM is switched out/paged in someway.
Yet another awesome video!!!
@Adrian's Digital Basement ][
In the BIOS, IIRC you need to disable the external cache on the board for the Pentium internal cache to work properly.
That floptical should work on ANY recent enough SCSI controller, so any SCSI II should be gr8
i have an unopened original imation box of 240MB floptical disks from 1997, i would send them over but they're not compatible with the 120 drives...
The various 486's had slightly different pinouts, some of which have to do with on-chip cache functionality. You'll have to check that your motherboard is set up correctly for the chip you're using.
Yes the floptical drives are dual-use.
It was the Samsung Galaxy S5 and Note 3 that have that micro usb 3.0
I’ve owned both phones at one point. Oddly enough most people thought you needed a 3.0 cable to charge the phone so the 3.0 connector was dropped from the next generation. I’ll never understand why people assume you can’t use a 2.0 cable on a 3.0 connector. Literally remember a customer inside Walmart asking for a 3.0 micro usb cable to charge a Samsung phone. At the time not only was that type of cable hard to find but but was generally double or even triple the price. Needless to say that customer went away angry as they couldn’t buy said cable because the only one they sold was out of stock. Oddly enough no amount talking to them was able to convince them that they could buy and use a standard micro usb to charge said phone
I can confirm Dot's Pretzels were available in Michigan when i was there last, at mainstream stores.
remember a review of a floptical drive in amiga format in the early 90s - 20mb I think. Never saw one in real life :)
14:35 Extra pins on Pentium Overdrive are mostly for power.
We have those Dot's pretzels here in NorCal.
The 486dx4-100 was tripled. There was an oddball 487sx that was socketed and could use an overdrive
I had this CPU and a similar motherboard, I don't remember the exact board, its been years, but I accidentally ran across a jumper on the motherboard that set the CPU multiplier, I removed the jumper, and the System Speed Doubled. I believe the bus clock divisor has to be disabled.
i had a 586 overclocked to 133Mhz. it ran cool also. Pretty straight forward. what about the overdrive.bat file?
Hi Adrian, the Pentium 1 have L1 cache have 16 KB (internal CPU), no L2 cache, and L3 cache have 128 KB found on your motherboard. The 80486 family CPU have L1 cache have 8 KB memory but few 80486 have 16 KB memory.
It was the Note5, I loved that thing, I Lost the Stylus while writing a Recipie down while working as a chef lol
Nobody ever told you this, but you could use a standard Micro-USB Cord!!!
It will still work 100% compatible with USB 2: I learned there somewhere around 2015. Blew My Mind that it wasn't common Knowledge. I had no idea lol
Those SK series chip appear to be equivalents to 74 series TTL chips, for example a 7447 is a BCD to 7 segment decoder, just like that 4547. Actually, I am wondering if the 4xxx means it's CMOS, so equivalent to a 74C47
Problem with L1 cache is definitelly somewhere on board. Inside 486 CPU you can only control cache disable (CD) bit and not write-trough (NW) bit in CR0 register. This can be probably enabled/checked by debug.exe. Rest of L1 cache control is in chipset. Also BIOS usually uses L1 cache as RAM before DRAM is detected and configured.
It seems the chipset can be either 82C596A or 82C597 and it seems to be cypress (cypress later bought contaq). There seems to be a datasheet for CY82C597 with registers description how to control cache (also L1 write back configuration).
For the micro USB3 connector the Samsung S5 had it and also the Note 3
A lot of mom and pop PC stores sold those A-Trend motherboards back in the day, IIRC they used Award BIOS mostly. Maybe a few used AMI.
50:11 hmmmm I'd like to see a repair of a tool you use all the time, investigating that cashe problem would make a GREAT video.
Check your Turbo switch
I have an external Maxell LS-120 for parallel port and one LS-120 disk. Years ago, i was able to make it work under Windows 98. Due to the parallel port connection, it is pretty slow. But what's interesting, it is able to format a regular 1.44MByte floppy to store ~32MByte. Cool device, but not very usable. These LS-120 floppies also seem to be very unobtanium. Another interesting aspect about that floppy disk is, you can see little squares on it. Not sure if that information is useful, but that's what i figured out, when i played with it.
I have a SATA hard drive just like that, and it still works to this day. I also find it weird that it has both molex and SATA for power.
Interesting video, and some very nice goodies.