Intel processor bug

Status
Not open for further replies.
Originally Posted By: itguy08
Originally Posted By: OVERKILL

I'm on Intel, I've had a few AMD systems over the years but the stability of the Intel chipsets back when VIA, ALI and their ilk were the main game in town for AMD really soured me on them.


I think a lot of that had to do with Windows rather than the VIA chipsets. Been using AMD since the 486 days and had K5, K6, K62, Duron, and now the Ryzen. The VIA chipsets were pretty solid but had driver issues with Windows. On my Duron box, the machine that pushed me to Apple because of 98 and ME issues all the stability issues went away when I put Linux on it. It became rock solid.


I had far more board failures (remember the rash of ABIT and ASUS VIA-based boards?) with VIA chipsets than anything else. ALi and SiS were arguably more stable in some respects, but their support was next to non-existent. SiS or ALi southbridge driver procurement was an exercise in itself. Makes me shudder just thinking about it, LOL

Quote:
Quote:
Growing up in the infancy of the personal computer, there were all kinds of brands that came, went or were absorbed. There used to a much broader array of graphics card/chip manufacturers: Trident, Cirrus Logic, 3DFX, VIA, ALi, SiS, Matrox, S3, Chips & Tech, OPTi, Oak, Real3D, Rendition, SGI, Hercules...etc.


I remember supporting that mess and those oddball drivers. It sucked...


It always made for an adventure finding drivers for stuff that was a few years old. I had quite the driver collection for that very reason
smile.gif


Quote:
Quote:
The CPU market was far less diverse, you had Intel, and then if you were cheap, you had AMD, VIA or Cyrix, all of which were a serious downgrade from a comparable Intel offering.


AMDs offerings were as good as or better than Intel. AMD actually was the 2nd source for 8086/8088 CPUs back in the day and their 286, 386, and 486 CPUs were as fast as if not faster than Intel's of the time. The K5 and K6 were faster in office type stuff but had a weaker FPU. The K62 aattempted to fix that but it's FPU was a little less. The Athlon was faster than Intel and was an awesome CPU.

Don't forget you owe the x86_64 instruction set to AMD who invented it and licensed to Intel. Intel was too busy with Itanium to try to extend X86.


The floating point performance was an issue for gaming (which was a big part of what I was supporting through that era) and AMD reversed that with the Athlon. However Coppermine had faster L2 cache, and that made it, in many instances the faster CPU. However Intel threw that all away for quite a stint on the detour that was the P4. Remember RAMBUS? I have some RIMM's here somewhere on my desk, LOL!

And yes, Intel's attempt to force the 64-bit game went absolutely nowhere, ultimately having to adopt AMD_64/x86_64. I bet that was a hard pill for them to swallow.

I find it somewhat ironic that the PIII (in its mobile form) that Intel abandoned in favour of the P4, would eventually evolve into the desktop CPU series (core) that put Intel seriously ahead of AMD performance-wise. That was a truly strange era to be building performance rigs
lol.gif


Quote:
Quote:
Chipset-wise, if you bought Intel, you got Intel unless you chose to intentionally cheap-out there as well. If you bought one of the others you got somewhat of a wide variety of potentially poorly supported garbage that finding drivers for was sometimes a serious epic.


Only past 486 was Intel in the chipset game. And even with the 486 and above there were other Vendors. IIRC VIA, SiS, and a few others did 486 and Pentium chipsets.

Aah the good old days - lots of good stuff and lots of junk....


Yes, other vendors made chipsets for Intel CPU's, but you'd have to be silly (or really cheap) to not just get the Intel chipset, which basically guaranteed reliability and support. That still holds true, they are still the best supported chipsets on the market, though their driver procurement process has become significantly more complicated than it used to be.

I used to have a Slot A Athlon 800 rig kicking around here somewhere, I may have to go looking for it. Had an AMD chipset on it. Was on an ABIT board that I re-capped. It worked REALLY well.
 
Originally Posted By: OVERKILL

I had far more board failures (remember the rash of ABIT and ASUS VIA-based boards?) with VIA chipsets than anything else. ALi and SiS were arguably more stable in some respects, but their support was next to non-existent. SiS or ALi southbridge driver procurement was an exercise in itself. Makes me shudder just thinking about it, LOL


I ended up settling on FIC for my AMD builds. IIRC it was the VA-503 for AT builds and their later ATX version. Like you I had nightmares finding some of the drivers for the oddball stuff you'd see. A lot of that came from the OEMs who would slap whatever was cheapest in them.

Quote:
It always made for an adventure finding drivers for stuff that was a few years old. I had quite the driver collection for that very reason
smile.gif


smile.gif
I remember getting my first CD burner in 98 or so - put tons of drivers on a CD and quit carrying 3.5" disks around!

Quote:
The floating point performance was an issue for gaming (which was a big part of what I was supporting through that era) and AMD reversed that with the Athlon. However Coppermine had faster L2 cache, and that made it, in many instances the faster CPU. However Intel threw that all away for quite a stint on the detour that was the P4. Remember RAMBUS? I have some RIMM's here somewhere on my desk, LOL!


Never was much of a gamer so the FP performance meant little to me. It wasn't till the K62 with 3DNOW that AMD got competitive in the gaming space. Remember the ___tshow that was the P4 - clock speed to the moon regardless if a P3 at a lower clock was actually faster. Got to hit those big #s.

Quote:
Yes, other vendors made chipsets for Intel CPU's, but you'd have to be silly (or really cheap) to not just get the Intel chipset, which basically guaranteed reliability and support. That still holds true, they are still the best supported chipsets on the market, though their driver procurement process has become significantly more complicated than it used to be.

PCChips sold TONS of motherboards!
smile.gif
 
Last edited:
Originally Posted By: itguy08


I ended up settling on FIC for my AMD builds. IIRC it was the VA-503 for AT builds and their later ATX version. Like you I had nightmares finding some of the drivers for the oddball stuff you'd see. A lot of that came from the OEMs who would slap whatever was cheapest in them.


Oh, that's EXACTLY what they'd do, sometimes you'd have these bizarre hybrids and you just knew that it was because that part was insanely cheap for some reason, not because it was the best choice
lol.gif


Originally Posted By: itguy08

smile.gif
I remember getting my first CD burner in 98 or so - put tons of drivers on a CD and quit carrying 3.5" disks around!


Oh the memories! I had a collection of 3.5" disks in a couple of those plastic flip-open 50 disk caddies! My first CD-ROM was a Creative Labs complete with its Sound Blaster 16 interface, LOL! There were what, like 3x different optical drive interfaces before IDE begrudgingly became the consumer standard?

Then of course the ability to WRITE discs came into the picture and it was SCSI or the highway! So many Adaptec cards... I still have an EXTERNAL 4x NEC CD-RW here, SCSI, that's left over from that era. RICOH made quite the drive back then, as did Plextor.


Originally Posted By: itguy08
Never was much of a gamer so the FP performance meant little to me. It wasn't till the K62 with 3DNOW that AMD got competitive in the gaming space. Remember the ___tshow that was the P4 - clock speed to the moon regardless if a P3 at a lower clock was actually faster. Got to hit those big #s.


Oh man, I did so many of those AT-swap K6-2 builds, because there was nothing Intel that would fit in those cases. So for some old boy with a Pentium 100 or 133, you could do a board/chip swap, keep the old PC100 or 133 that was in there (unless he was on EDO) and UPGRADE! It was what, the 450? That sound right? That was the "bargain" in that series.

Originally Posted By: itguy08

PCChips sold TONS of motherboards!
smile.gif



And now you've given me nightmares. The pinnacle of garbage in a MicroATX form factor: PCChips! All of the cheapest possible components slapped on a board with the bottom of the barrel capacitors and generally coupled with a Celeron or Duron :shudder:
 
Originally Posted By: OVERKILL
However Coppermine had faster L2 cache

I remember Coppermines and their cousin the Tualatin with 512kb L2 cache. Built dual-370 socket board iWill brand, first and last time hearing about them) with a couple of PIII-S Tualatin 1.26GHz and I loved that machine. Underclocked, undervolted it and ran it fanless with laptop drives in a RAID. Back then a virtually silent desktop was pretty rare and I loved it, until the board got physically damaged. Thanks for triggering the memories
 
Originally Posted By: PeterPolyol
Originally Posted By: OVERKILL
However Coppermine had faster L2 cache

I remember Coppermines and their cousin the Tualatin with 512kb L2 cache. Built dual-370 socket board iWill brand, first and last time hearing about them) with a couple of PIII-S Tualatin 1.26GHz and I loved that machine. Underclocked, undervolted it and ran it fanless with laptop drives in a RAID. Back then a virtually silent desktop was pretty rare and I loved it, until the board got physically damaged. Thanks for triggering the memories


Do you remember the ABIT BP6 and later VP6? Dual boards, I built a number of systems on them, the OC'd Celerons were the CPU of choice for the former.
 
[censored]
[I literally can't put enough censored words here. Just imagine, like 80 of them]
Might be time for me to look into a minimum-viable upgrade on my old beast. Motherboard/CPU/RAM. My poor old X58 i7 system has worked spectacularly for 8 years now. Originally built 01/2010 with a 920, then I upgraded to a 980X. It still [censored] amazes me its so powerful and usable so far into its life. I typically run a Plex and UniFi servers on it, along with a W10 VM, 3 or 4 chat programs, Acrobat with like 20 tabs, and two browser windows with 20 tabs each. Then I still have the resources to do some light gaming or video editing. 20 or 30% performance impact from the software mitigation would totally destroy it.
Ugh, Im screwed.
 
Originally Posted By: OVERKILL
Originally Posted By: PeterPolyol
Originally Posted By: OVERKILL
However Coppermine had faster L2 cache

I remember Coppermines and their cousin the Tualatin with 512kb L2 cache. Built dual-370 socket board iWill brand, first and last time hearing about them) with a couple of PIII-S Tualatin 1.26GHz and I loved that machine. Underclocked, undervolted it and ran it fanless with laptop drives in a RAID. Back then a virtually silent desktop was pretty rare and I loved it, until the board got physically damaged. Thanks for triggering the memories


Do you remember the ABIT BP6 and later VP6? Dual boards, I built a number of systems on them, the OC'd Celerons were the CPU of choice for the former.

Oh yeah! Actually bought a BP6 second hand when I lost the iWill DVD266 board, turns out it didn't support the Tualatins. I remember the love for that board and the BX chipset in general though, for sure
 
Originally Posted By: PeterPolyol
Originally Posted By: OVERKILL
Originally Posted By: PeterPolyol
Originally Posted By: OVERKILL
However Coppermine had faster L2 cache

I remember Coppermines and their cousin the Tualatin with 512kb L2 cache. Built dual-370 socket board iWill brand, first and last time hearing about them) with a couple of PIII-S Tualatin 1.26GHz and I loved that machine. Underclocked, undervolted it and ran it fanless with laptop drives in a RAID. Back then a virtually silent desktop was pretty rare and I loved it, until the board got physically damaged. Thanks for triggering the memories


Do you remember the ABIT BP6 and later VP6? Dual boards, I built a number of systems on them, the OC'd Celerons were the CPU of choice for the former.

Oh yeah! Actually bought a BP6 second hand when I lost the iWill DVD266 board, turns out it didn't support the Tualatins. I remember the love for that board and the BX chipset in general though, for sure


grin.gif
 
So maybe I missed it, but is the proposed patch/fix something Microsoft is going to include in a Windows Update for the Windows crowd? And will it really decrease processor performance?
 
Processor performance should never determine yes or no on keeping the computer Windows Update up-to-date using Windows 10.

ALWAYS keep Windows Update turned on and ready to go. Ways to deal with performance issues is expanding or enlarging the system. Only other alternative is change operating systems.

I find Firefox easier to manuever around, operate and clean, versus that of Chrome. A little slower Firefox would never get me to change to Chrome or another system like Ubuntu.
 
Last edited:
Latest news announced by Google, Intel, and others is that this problem is beyond the scope of just Intel CPU's and there are at least two different flaws that can be exploited. While AMD is not affected by the first one, it, and ARM are by the 2nd.

CBC article can be found here.

Originally Posted By: CBC

What are these updates trying to fix?

The two flaws let attackers access parts of a computer's memory that they shouldn't normally have access to, by abusing the way that computer processors are designed to handle information more quickly.

One of the flaws, called Meltdown, allowed the researchers to access data stored in the kernel — the core of a computer's operating system, which runs in a protected part of a computer's memory, and effectively watches over everything your computer does.

By design, applications can't access the kernel, a protection that's built into the hardware of the CPU itself. But the researchers found a way around that, giving them access to the kernel and, from there, any data stored in a computer's memory — which could include everything from passwords to photos. This attack has only been found to work on processors made by Intel. "The bug basically melts security boundaries which are normally enforced by the hardware," the researchers wrote.

The other flaw, called Spectre, allowed researchers to target data that applications store in a computer's memory directly (typically, applications can't access the memory used by other applications).

It's related to Meltdown, but differs in a number of ways that the researchers detail in a pair of technical papers. This attack was found to work on Intel, AMD, and even ARM processors, which are commonly used in mobile devices such as smartphones and tablets.


Regarding mitigation:

Originally Posted By: CBC

While Meltdown can be patched, Spectre will be much more difficult to defend against long-term because of the way that CPUs are designed — and that's worrying, because it's the vulnerability that affects a far wider range of chips.

The researchers say any Spectre-specific software patches for applications, operating systems or CPUs should be considered stopgaps while more research takes place.

"As it is not easy to fix, it will haunt us for quite some time," the researchers wrote.
 
Originally Posted By: OVERKILL
CBC article can be found here.

The article mentions:
Quote:
Microsoft pushed out an automatic update for Windows users Wednesday night.

I'm looking through the Windows Update history on my Win 7 machine and don't see any updates being installed in the last few weeks.
 
Originally Posted By: Quattro Pete
Originally Posted By: OVERKILL
CBC article can be found here.

The article mentions:
Quote:
Microsoft pushed out an automatic update for Windows users Wednesday night.

I'm looking through the Windows Update history on my Win 7 machine and don't see any updates being installed in the last few weeks.



Just an FYI, but KB4056894 just showed up for my Windows 7 VM (2018-01 Security Monthly Quarterly Rollup).
 
Originally Posted By: OVERKILL
Just an FYI, but KB4056894 just showed up for my Windows 7 VM (2018-01 Security Monthly Quarterly Rollup).

I just did a search for new updates on mine. It did not find anything. Strange.
 
Originally Posted By: OVERKILL
Just an FYI, but KB4056894 just showed up for my Windows 7 VM (2018-01 Security Monthly Quarterly Rollup).


Yep, I loaded that update yesterday. It was big, around 280 MB if I recall correctly. Any way to know if it patched a fix for this processor issue?

No mention of it in KB4056894, so I'd assume not.
https://support.microsoft.com/en-us/help/4056894/windows-7-update-kb4056894

Update: Looks like Windows Update KB4056894 does address part of the vulnerability: LINK
 
Originally Posted By: itguy08
I think a lot of that had to do with Windows rather than the VIA chipsets.


Nah. In the early 2000s I worked for a company that built PC add-in cards, and the code in the driver had something like:

Code:
if ( VIA chipset ) then

switch to slowest possible PCI mode


Because they were so buggy and it wasn't worth the effort to try to find settings that worked properly while maintaining performance.

So odds were that if you plugged one of our cards into your VIA-based PC it would work fine, but performance would be rather less than if it had been plugged into an Intel or AMD chipset.
 
Status
Not open for further replies.
Back
Top