Image Map Image Map
Page 4 of 10 FirstFirst 12345678 ... LastLast
Results 31 to 40 of 97

Thread: Worst x86 CPUs over the years

  1. Default

    Quote Originally Posted by Eudimorphodon View Post
    The Via C3/C7 are pretty overall terrible
    C7 had better performance at the given power consumption level than the pentium4. Its was a fine cpu, i used it for a while.
    The performance of VIA C3 is very similar to the early pentium1 chips on the same clock.

    Both was very good cpu for its age - the bad reuptation is due that some corporations used the C3 chip even after 2010 and some tried ti use the C7 around 2013, obviously a decade after its intended usage.

  2. #32
    Join Date
    Dec 2012
    Location
    Portland Oregon
    Posts
    128

    Default

    The only real advantage/selling point to the C3 was it's TDP. Back in the day, Socket 370 chips were getting better, stronger, faster at the expense of Wattage. When I was still at Intel in 1999, I remember having a conversation with a CPU engineer who, when asked about rising Wattage, replied that the market would find a cooling solution. In fairness, this was when few people had/used laptops/portable computing. My guess is that VIA saw an opportunity to focus on TDP vs speed. I have my VIA C3 still in my collection. At the time, it worked well from a compatibility standpoint (i586) with really good thermal qualities. Speed/performance.... not so good.

  3. #33
    Join Date
    Dec 2012
    Location
    Portland Oregon
    Posts
    128

    Default

    Quote Originally Posted by Chuck(G) View Post
    I've got an 820 system with 1GB of RAM. What impresses me is how hot the bloody things run.
    Indeed. I have a Dell Precision 220 workstation in my collection. Socket 423 P4 with 512 Meg RDRAM. Use it/Mesure it side by side on NT4/WinXP vs a SDRAM system and there is no competition. Even today, its runs really fast for its age. But it comes with many fans to keep it cool.

  4. #34
    Join Date
    May 2011
    Location
    Outer Mongolia
    Posts
    1,931

    Default

    Quote Originally Posted by Geri View Post
    C7 had better performance at the given power consumption level than the pentium4.
    I think the reason the C7 particularly gets under my skin is because my first exposure to it was in an HP 2133 notebook, which came out within a couple months of the first batch of Atom 270 netbooks. And the Atom was a much, *much* better CPU by every measure (including TDP, by a fairly large margin). Maybe it's dirty pool to compare them directly but both were still being sold new in 2008.

    An article I read at one point claimed that HP had been hoping to release the 2133 with the Via Nano instead, but it ended up running so late they stuffed a C7 on the board and booted it out the door anyway. I've always been kind of curious about the Nano, but it doesn't look like it ever really shipped on anything even remotely mainstream. They made a few ITX motherboards, and it looks like there was at least one no-name OEM netbook in limited circulation that appears to have combined the Nano with the same hateful S3 Chrome graphics chip that you found in most C7 machines, but I've never seen one in the flesh.

    I will admit the C3 was perfectly adequate for the things it was mostly used for, IE, Citrix terminals and whatnot. There's always a place for CPUs like that. (Granted these days that niche belongs almost entirely to ARM.) But as a mainstream CPU after about 2002? Heck no.

    FWIW, Centaur, the outfit Via acquired that was responsible for the C3/C7 architecture (and the Nano), was also responsible for the aforementioned WinChip. To me it's kind of a shame that Cyrix's technological bloodline ended so ignominiously after the Via acquisition because despite a few black eyes the company did have a quite a few "creative", if not necessarily great, CPUs under their belts. The Cyrix Fasmath coprocessors for the 386 were very fast (kind of ironic given how floating point became Cyrix's Achilles heel), and while the 486SLC and DLC were obviously inferior to a real 486 their performance *could* be pretty amazing considering the limitations they were dealing with.

    (Although maybe the 486SLC deserves a mention on a list of worst x86 CPUs as well, given the absurdity of trying to combine a "486" name with a 386sx socket. Although to be fair then we also need to ding the IBM 486SLC2, which actually performs amazingly well given its 16 bit data path, but is shamefully hobbled with that 24 bit address buss.)
    Last edited by Eudimorphodon; January 22nd, 2020 at 11:10 AM.
    My Retro-computing YouTube Channel (updates... eventually?): Paleozoic PCs

  5. #35

    Default

    Quote Originally Posted by Eudimorphodon View Post
    (Although maybe the 486SLC deserves a mention on a list of worst x86 CPUs as well, given the absurdity of trying to combine a "486" name with a 386sx socket. Although to be fair then we also need to ding the IBM 486SLC2, which actually performs amazingly well given its 16 bit data path, but is shamefully hobbled with that 24 bit address buss.)
    Also reserve a spot for the IBM 486SLC3, which is actually internally a fully 32-bit 486DLC3 (a.k.a. "Blue Lighting", 486BL, or 486BLX) which was crippled down to the SLC's 16-bit external data bus. I have one in my PS/2 Model 56. At 60 MHz, it's only about as fast as a 486DX-33, but for a system that originally came with a 386SX-20, that's actually pretty darn good.

    Speaking of which, the IBM Blue Lightning was the first x86 CPU to reach 100 MHz -- a big deal at the time, because it suddenly made all cases with two-digit LED speed displays obsolete. (I remember many computer show vendors configuring them to show "99" or just "HI".)

  6. #36
    Join Date
    Apr 2015
    Location
    Austin, Texas
    Posts
    1,800

    Default

    Quote Originally Posted by gepooljr View Post
    The only real advantage/selling point to the C3 was it's TDP. Back in the day, Socket 370 chips were getting better, stronger, faster at the expense of Wattage. When I was still at Intel in 1999, I remember having a conversation with a CPU engineer who, when asked about rising Wattage, replied that the market would find a cooling solution. In fairness, this was when few people had/used laptops/portable computing. My guess is that VIA saw an opportunity to focus on TDP vs speed. I have my VIA C3 still in my collection. At the time, it worked well from a compatibility standpoint (i586) with really good thermal qualities. Speed/performance.... not so good.
    VIA focusing on TDP rather than performance is what really sank them, and later Transmeta. In both cases, the public really didn't really see the need to worry about power consumption because mobile devices weren't really a thing yet. To the public, low power consumption devices was equivalent to sacrificing performance and seen as a negative thing. It wasn't until at least a decade later when mobile devices became more important and powerful enough that people started to see the benefits of low power devices. I mean you did have things like PDAs and mobile phones, but they weren't seen as general purpose mobile computing devices at the time.

    I still remember Transmeta trying to gen up public interest by having demos of how little power their CPUs used compared to the average x86 processor and nobody really cared that much. It was an interesting curiosity at the time, but the performance of their CPU was so abysmal that any initial interest was quickly lost. While Transmeta claimed their code morphing emulation engine was as good as a bog standard x86 CPU, it was far from it. Many applications wouldn't run at all, or have terrible performance issues. The code morphing software itself also used up a huge amount of system RAM, I think something like 64-128 MB, further impacting the useful scenarios of their CPUs.

    Another feature claimed by Transmeta is that their CPUs would never become obsolete as newer CPU instructions were released, like SSE2, SSE3, etc. because they could just update their code morphing software. This turned out to never materialize, mostly because they went bankrupt in 2009.

  7. #37
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    33,391
    Blog Entries
    18

    Default

    I've seen words about the C3+ chips having an alternate instruction set. Looks interesting, but never had the urge to test it out.

    AIS info
    Last edited by Chuck(G); January 22nd, 2020 at 10:38 PM.

  8. #38

    Default

    Quote Originally Posted by jasa1063 View Post
    The interesting thing is Cyrix and VIA ... I have no joy when I have to work on those things because they are so slow.
    I use Athena thin clients with VIA CPUs to emulate the old Commodore HDD. Not just the HDD but also FDDs using images. At the same time. The system runs W98 DOS when I need real time operation, XP and higher won't allow direct I/O access, and it runs XP when I need to exchange images with other computers over the local network or USB.
    With kind regards / met vriendelijke groet, Ruud Baltissen

    www.baltissen.org

  9. #39

    Default

    Quote Originally Posted by jasa1063 View Post
    Although it really started the entire PC industry with the IBM PC, the 8088 was not high on my list. .... Some took it as a challenge, I took it as chore.
    See it as the Volkswagen Beetle: very reliable and it ran and ran and ran.... But just like because my kids got bigger and therefore I had to get a bigger car, I was forced to run Oracle V6 and I needed at least a 386. So I entered the 486 era.
    With kind regards / met vriendelijke groet, Ruud Baltissen

    www.baltissen.org

  10. #40
    Join Date
    Aug 2006
    Location
    Chicagoland, Illinois, USA
    Posts
    6,294
    Blog Entries
    1

    Default

    I've had a love/hate relationship with the 386sx. I have very fond memories of the 386sx-16 era, where it was a very cheap way to get a flat 32-bit address space, protected memory, hardware multitasking, etc. in 1990. I have hateful memories of the latter 25/33-MHz era, as the 16-bit memory access really threw people off and led to disappointment (you'd try to run a program with "386/33" as a minimum spec and it would be unacceptably slow because "386/33" assumed a 32-bit memory path).
    Offering a bounty for:
    - A working Sanyo MBC-775, Olivetti M24, or Logabax 1600
    - Music Construction Set, IBM Music Feature edition (has red sticker on front stating IBM Music Feature)

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •