Intel vs AMD vs Cyrix DX2-66 Face Off

[@Evansmustard]

These videos are so awesome dude. Great idea on this!

[@AncientElectronics]

thank you.

[@Daehawk]

From 1994 - 1998 or so I had a intel 486SX33, AMD 486DX4-100, and a Cyrix P166+

[@philscomputerlab]

Just started watching and don't know the outcome yet, but the last time I looked into this I took an Intel DX4 100, not sure if it is WT or WB, and the top AMD with 133 MHz. I clocked the AMD also at 100 MHz and the Intel was a tiny bit faster. I admit that whole cache stuff is confusing as heck, I doubt it matters much really.

[@AncientElectronics]

I did that as well and had the same result, though not in this video. I've never had WB enhanced CPU's run correctly for me though so I cant speak for how much of a boost it gives. in my WB experiment in this video I actually got worse performance though I'm unsure why.

[@derbiber842]

Phil srsly, I just want one buck only everytime you mention that the intel dx4 is faster clock by clock than the amd one on any forum or youtube. I swear I never ever have to think about working again xD

[@philscomputerlab]

derbiber842 Haha, don't worry, I''ll do a video at some point :D

[@derbiber842]

not another intel movie :O ;) then please put the cyrix 5x86 in there as well

[@philscomputerlab]

derbiber842 That's not a real 486 in my books. And there is the issue that one needs to enable things with tools to make it go faster, but apparently this is not 100% stable? Still an interesting chip, I agree.

[@adgarza]

Interestly enough, I was an user of OS/2 since version 2.1. I was really in love with such an operating system and I am an AMD user since back in 1985. As the OS/2 Kernel is far more complicated than the MS-DOS one, it needs to load some libraries corresponding to each processor. I installed OS/2 in Am386, Am486, Am586 and, even, in AMD K5. I had a Cyrix 6x86 and I installed OS/2 Warp on it successfully. I think if you do an installation of OS/2 in an Intel processor, the system will load the libraries for Intel processors. If you do the setup in AMD or Cyrix processors, it will include the libraries for such processors. Is, more or less, the same with WinNT.

[@vwestlife]

Did you try the Cyrix with write-back cache enabled? It's supposed to support it: www.fermimn.gov.it/inform/materiali/evarchi/cyrix.dir/486-tb.htm I had a Cyrix 486DX2-80 with that same green heatsink on it and although it was generally an underwhelming CPU (not much faster than a DX2-66) it did definitely support write-back L1 cache.

[@AncientElectronics]

no I did not. I thought only the DX2-66's with the "write back" logo on the heatsink were the enhanced chips. I'll give it a try.

[@thudtheace]

BTW The Cyrix 486 chips do support write-back cache as well.

[@tahustvedt]

15% slower in Doom for the Cyrix is more of a difference than I expected.

[@jeffhatfield2230]

with all of these chips you have to set the jumpers correct for them to run at optimum speed and to also run any os2,3,4 sysy. I am 57 yrs old and started building systems in the late 70,s as in heathkit and comodore pet systems. so relize that you had to know the boards and jumpers on a lot of different boards!

[@AmstradExin]

I have to say that may be true. Most 486 boards have over 20 jumpers to set for all kinds of things. It was a pain in the ass for me to get a Pentium overdrive working CORRECTLY....

[@jeffhatfield2230]

yea i know what ya mean but sfter building 500 486's for an independant local company i definatelyt know whatya mean

[@kanopus06]

As long as I remember, any 486 faster than 50 MHz required a heatsink at least. I did run my Intel DX4 with a small heatsink and fan to cool it down. Maybe this is the cause why the AMD DX2 failed : it burt itself.

[@AncientElectronics]

It's a possibility but I kinda doubt it. I've known people that have run dx2-66mhz chips for years with no heatsink or fan and had no issues. I'm not saying you shouldnt. I always advise cooling even on chips it's not really required but again, I left them off for this video so viewers could see the chip.

[@-GameHacKeR-]

@@AncientElectronics that was a bad idea

[@AncientElectronics]

@@-GameHacKeR- When I made this video I quite literally had a small bag of DX2 486 chips so I wasn't exactly heart broken. For this video I just wasn't all that concerned so I did things haphazardly.

[@amdintelxsniperx]

also remember this is very board dependant 486s really shine on specific motherboards and things will always very from board to board

[@AncientElectronics]

very true, this board gives a little better results then my intel dx2-66 running in my main rig. so many variables from cache timings and speed to chipset performance.

[@amdintelxsniperx]

exactly my old chipset i had with a umc 386 was keeping pace with many 486s i switch to an evergreen 586 and it was actually a few frames faster then my friends setup he had the same cpu as the one on this video

[@HappyBeezerStudios]

That even holds true later on. When I moved my Pentium III (I know not exactly 486 time) from a VIA chipset to a i815 the performance increased about 15-20%. On the same clockspeed and configuration!

[@jbolleman74]

Please heatsink your CPU's before you kill them all! NOTE: the DX4 cpu's must be set to 3.3volt

[@AncientElectronics]

My board has jumpers to set voltage and I indeed did switch back and forth depending on the appropriate CPU else my DX4's would of gotten fried at 5V. as for the heatsink/fans I did not have one on the CPU so the viewers could see the CPU actually being tested. These CPU's run cool enough that its pretty safe to go without cooling for short periods of time. Also I had a heatsink/fan that was placed on the CPU when I wasn't filming them so the time they went exposed was actually very short. This wasn't my first rodeo.

[@jbolleman74]

Good to hear. I benched a lot the days of these CPU's. Had practicly all types from SX25 to 133 rated speeds. Really nice memories overclocking and benching the crap out of them. Luckily there is still a 486 160mhz in my basement. Thanks for your video.

[@AncientElectronics]

There's something oddly enjoyable about benching CPU's. especially fiddling with BIOS settings and memory timings and then running the benchs again and the anticipation of the possible new outcome. I love the AMD 5x86 133 @ 160mhz.

[@Daehawk]

Those old chips only had heatsinks back in the day.

[@Baoran]

I have also noticed that old hardware is less tolerant to temperature changes. Even if a cpu was fine without heatsink and fan 25 years ago, temperature changes do kill old hardware more easily..

[@theKattoz]

Hey im pretty sure that Dx4 needs a heat sink and a fan ?

[@AncientElectronics]

Oh, that's certainly a good idea for the Dx4 chips. I left mine off for this video since it was only for a short period of time and to show what I had installed. I did actually have a fan blowing on them when I wasn't recording though.

[@theKattoz]

I think its also written on the chips that they req heatsink i just to play alot around with 486´s but i sold most of what i had.

[@matthewday7565]

Just pooped up in my feed... often wondered if the WB L1 cache was a good combination with the WB L2, maybe WB L1 with WT L2 might fare better - with both in WB, there may be some weird delay as the WB L1 hands off to WB L2, as it can either be handing off due to cache line replacement (L1 to L2, remains WB in L2), or to external snoop (forcing an effective writethrough of the L2)

[@Kenny-bw2cz]

The most interesting thing to me would be those cpu information tools. Where can I find that?

[@Kenny-bw2cz]

I think I found it.web.inter.nl.net/hcc/J.Steunebrink/chkcpu.htm

[@magnum333]

Great stuff. I have a Am486 DXL2-66. What would be the difference with a Am486 DX2-66?

[@AncientElectronics]

as far as I can see from doing a search, performance wise the dxl2-66 is exactly the same as the dx2-66

[@magnum333]

Thanks for searching. I found the same. Still I wonder, what does the L stand for. As a side note, I got it working with 3.3V, although on the internet I found it requires 5V (cpu-collection dot de). I wasn't sure and wanted to err on the safe side, and it worked.

[@WaybackTECH]

The 486 era was interesting for AMD. The AM386 is a 100% clone of the Intel 386 chip, AMD shared a licensing agreement with Intel which allowed such cloning, and performs identical in all tests. Intel got butt hurt that AMD was going to release their 386 before Intel, and kept them tied up in court till around 90 or 91 something like that, not the mention the i386 chips were riddled with bugs. AMD still though had the designs for the Intel 486 in their hands, since they still held a licensing agreement with Intel in the early 90's, but the 486 was a modified version to get around Intel suing them over patent infringements. AMD didn't actually design a chip from scratch till the K5, and the K6 is really just a tweaked and modified NX. Cyrix chips on the other hand were designed 100% in house with a small team in Texas who reverse engineered Intel cpu's and never shared a licensing agreement with Intel, and as such their instruction sets were not 100% compatible with Intel, though close enough to get the job done most of the time.

[@HappyBeezerStudios]

Correct, AMD's 486 were direct clones. Cyrix designed their own. The design of Cyrix was also used by IBM, TI and ST. AMD owned Cyrix later for some time until they sold them to VIA. And VIA is actually still in the market.

[@ajax700]

@WaybackTECH​ (cyrix) "their instruction sets were not 100% compatible with Intel" ? Poor dead cyrix, the only sin they comitted is to sell the 5x86 not advising it was NOT a pentium class processor. It wasnt, it only had 486 instructions. But had some pentium like architecture features, and most important: performance. cyrix's cpu had the exact same instructions as intel and amd, 100% compatible. They would have not sold even a single cpu to ibm, ti, compaq, if they weren't. "AMD didn't actually design a chip from scratch till the K5" The K5 was a redesigned amd Am29000. No redesign from scratch.

[@theshermany]

These chips need heatsinks and fans preferably. I even trued the first peltier coolers on 486s. I remember having a miro video card for pci slot on a umc chipset board.

[@infinitecanadian]

Do math coprocessors help with games that have a pseudo-3d environment (like Wing Commander or Red Baron)?

[@AncientElectronics]

As far as I know. No. I assume they could but the games are not programed to take advantage of the FPU. Very few games prior to the Pentium era seem to of been.

[@jozsiolah1435]

Intel stands for intelligence, you can activate it by reading, plain text games, or Monkey Island text game. It turns on an autopilot, so if you shoot or drive, it corrects your mistakes. Stunts also turns it on hen you choose menus, select cars, colours, etc. AMD stands for Agility, Magic, Dexterity. Better chance for finding rare websites.

[@jameslewis2635]

The results are pretty much what I expected to see for the CPU's on test. At the time the AMD chips were generally regarded as 'cheap chips' and were mostly unknown although I heard later on they were a match for Intel in this day because they were the same design. Cyrix however has always had a reputation for underperforming against the competition at similar speeds. Thinking back I can't think of a single time they held the performance crown. Generally the only thing Cyrix sold that were good and unique from their competition was the way they allowed weird upgrade paths from sockets that were considered dead such as the overdrive chips for 386 and 486 boards.

[@michalzustak8846]

James Lewis Cyrix 5x86 is faster than any other 486 chip.

[@WolfTheDog]

Cyrix are just better for calculations, not so much average user use. I think this is why it benched lower in games. Nice of them to include a heat sink though. Some 486s run hot

[@zach446]

Or the clock before the multiplier. Intel was 25mhz and amd was 33. People call the amd dx4 the dx3 for that reason

[@nopochoclos]

11:26 please use a HEATSINK!!!! and remove these stickers, a lot i say a LOT! or processors are dead because these stickers on 90s...

[@CobraTheSpacePirate]

that era of the 486s...AMD had the exact same IC die as the intel one...becuase of licensing and stuff...internally, they were exactly the same chip...I think that the AMD was the manufacturing was making the Intel chips... the factories...they had the same exact die!

[@Baoran]

I thought that pentiums were the first cpus that really needed extra cooling? Also I don't think you should disqualify a chip just because results are not what you expected.

[@AncientElectronics]

the later 486 CPU's generally require extra colling. you could probably get away without with a 66mhz if the case is well ventilated....maybe but its not generally advised. some of the 486's above 66mhz say right on them requires heatsink and or fan. I actually got scolded several times in the comments for not using a heatsink/fan. As for the 486 write-back that I disqualified. It wasn't that it wasn't what I expected so much it's that I was fairly certain the chip wasn't working properly. I couldn't verify the write-back cache was actually running as write-back.

[@Baoran]

What made you think it wasn't working properly other than it being slower than you expected?

[@AncientElectronics]

All the research I did indicated that a CPU with write back should be generally faster in most situations at least by a small amount. If you have two cars of the exact same make and one has a V4 and the other a V8 if you get the same results or worse with the V8 you can fairly safely assume the V8 is not functioning correctly. I'm not a "car guy" though so I hope that's a manageable analogy. Now it is also quite possible that simply none of my benchmarks were programs that utilized the write-back and thus ran...worse? BUT being as I did not have a program that I can say confidently verified the write-back was working I decided it was better to simply leave the results out rather then possibly add faulty results, adding the results from a Intel DX2 write-back CPU here was rather a bonus thing anyways so it wasn't really detracting. Now if you can suggest either a program that can verify write-back is being used on the CPU or a game/games or program that are known to run better with write-back enabled I can use that to rerun the tests and revisit this.

[@Baoran]

if you researched the subject and accepted the result, there would have been no point in testing it. If you test something, there needs to be a chance for a different result than the results you read when you researched it. If it wasn't faster because of the motherboard, that is also a result because many people might have similar motherboards where it doesn't matter which version of the cpu you have. Is there a reason why you didn't believe the program that showed it was using write back cache mode after you did set up the jumpers?

[@AncientElectronics]

Again, it was a BONUS.I did not want to post the results if there was a chance they were faulty. This was based on the research that the write-back SHOULD be faster. The point of the test would purely be to see how much faster. the program I used to check was cupid in DOS. again, an analogy would be running a v4 engine...then adding a turbo and getting no speed increase, its a good bet the turbo is broke. I did not want to add results that could be wrong and could be confusing if there was a chance they were faulty. It was irrelevant anyways to the main point of the video. It could also be the board wasn't configured properly either somewhere in BIOS or a jumper for write-back but it wasn't the point so I didn't spend any time to figure it out. If I want to be sure and revisit this I will buy another enhanced dx2 in the future and run more tests on many motherboards with more benchmarks called something like "write-through DX2 vs write-back DX2" but for this particular video I did not feel confident using the results nor would they be relevant to the point of the video as being the stock speed difference between a Cyrix, AMD and Intel DX2.

[@-GameHacKeR-]

You should watch adoredtv's intel anti competative video, He goes over all intel's legal battle with amd back in the 90's but a quick sum up is intel hired amd to produce cpus as a out source to meet demand, that included a deal for amd to use intel's x86 instruction set for so many years, but before the deal ran out intel pulled out of the deal. Anyway watch the video you'll enjoy it. Also you will need to get rid of that battery before it leaks and eats through the mainboard. Should be plenty of guides online to remove them and replace with a button cell battery case. Just make sure to locate the +&- places.

[@christopherdwyer7570]

Go woodgrain or go home!

[@-GameHacKeR-]

Also you do notice that both those AMD chips say, HEATSINK AND! FAN Required, And your not using a HEATSINK or FAN on ether the AMD chips or the Intel chips, So yeah it maybe died. or I should say you maybe killed it. :(

[@pipschannel1222]

I was going to say the same thing. Even my Intel DX2-66 gets about 80 degrees celcius when I feed it the onions.. But maybe it's just me; I even run a heatsink on my 386DX-25 ;-)

[@theatheisthammer]

And made a 150mhz I think I could be wrong though

[@BandanazX]

Cyrix Instead (TM)

[@zach446]

I believe amd chips had faster front side busses

[@ifrit05]

Lol I have both of these laying in a CPU bin.

[@martijnvanzanen4075]

Nice video and respect for you dude. But, aren'/t you a kinda of copy cat on youtube? I know U won't answer me. but myself and many wonder. *and I won't call names"

[@AncientElectronics]

yes, I scan youtube daily and when I find a video about old computers I create a cheap knock off. I'm so ashamed I've been found out. I do hope this doesn't effect my revenue.

[@kubamarciniak8829]

guci gan