-
We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.
Intel does it again
Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
VirtualLarry
No Lifer
Good point.. Via should have pretty-much recalled... uhm... ALL of their SocketA chipsets. (Maybe with the possible exception of their 8237 southbridge, I haven't heard too much grief about it.)
VirtualLarry
No Lifer
I thought that ATI had a bus license. (Actually, doesn't the P-M use the same bus as the P4 does? Or are there differences because it's a mobile chip? I know that the basics are the same, QDR and all that.)
Btw, the change to PCI latency settings was just a workaround, the true issue was that Via never implemented "bus park" functionality in their PCI bus, because the actual spec listed it as an "optional" part, even though it was de-facto required for proper operation by most PCI cards on the market. Intel's designs had always supported it, and the cards were designed to use it. Via took the cheap/easy way out, and had severe compatibility/performance problems because of it.
StrangerGuy
Diamond Member
Oh really? The KT266A and K8T800 Pro rocked IMO. 😛
White Widow
Senior member
Yes, shame on Intel for makingn such a glaring mistake. Compared with their competitors, they have by far the largest budget in development and testing. The fact that this got by them is supremely ebarassing.
However, at least they are doing the right hting and recalling all the problem chipsets, rather than telling people to wait for a BIOS upgrade, or to just buy some other RAM. Virtually every other PC hardware company that has problems like this never recalls their equipment, nevr mind admitting it. You have to gove Intel creit for sucking it up and admitting the problem, AND going to huge expense to make it right. You don't see AMD recalling all their Opterons...
However, at least they are doing the right hting and recalling all the problem chipsets, rather than telling people to wait for a BIOS upgrade, or to just buy some other RAM. Virtually every other PC hardware company that has problems like this never recalls their equipment, nevr mind admitting it. You have to gove Intel creit for sucking it up and admitting the problem, AND going to huge expense to make it right. You don't see AMD recalling all their Opterons...
I had a KT266A. I can't really give an opinion on it since the motherboard was one of the ones with crappy capacitors that caused frequent crashing (Gigabyte 7VTXE). The previous chipset, the KT266, had a lot of problems. The chipset before that, KT133, has really sucky driver support; the drivers don't help to make the AGP bus work properly.
PhlashFoto
Diamond Member
"Should have known better??" the chipset is probably fine, just a small error in manufactoring. Big difference between architecture/design flaw and a manufacturing flaw.
I could careless how small a company is, Via should have correct or some how resolve the problem with the chipset, instead of telling its customers, "Oh well, we'll fix it next time. Buy the new product."
AWhackWhiteBoy
Golden Member
This was when it first came out,i'm not sure if they are licensing people now. I know Via had chipsets ready to go and Intel refused to give them the license to use them.
As for the PCI issue, I looked it up again for the first time in a long time.
"High PCI-traffic due to the use of sound-, tv-, videocapturing-cards or NICs, causes data loss on the VIA 686B southbridge which is used with the VIA KT133A/KT133 chipset. The problem will occur especially on computers using a Creative Labs Soundblaster Live! soundcard. When the CPU waits for 3 busmaster requests until it accesses the bus itself a buffer might overflow and data will be lost. VIA offers a patch that will allow the CPU to wait only for 2 requests preventing the overflow and data loss. This patch might affect audio- and videostreams which will have dropouts due to the minimized bandwith. Since the IDE-controller of the 686B is affected by the data loss it is a serious problem. It is rumoured that a new revision of the chipset has been released to fix the problem.
DID ANYONE ACTUALLY READ THE LINKED ARTICLE???
The problem is not with the Intel chipsets - it's a problem with the memory modules.
There seems to be a problem with the reading comprehension skills of some people on these forums. Perhaps they should be recalled to elementary school for remedial English. :|
The problem is not with the Intel chipsets - it's a problem with the memory modules.
There seems to be a problem with the reading comprehension skills of some people on these forums. Perhaps they should be recalled to elementary school for remedial English. :|
AWhackWhiteBoy
Golden Member
"The modules cause problems when used in notebooks with certain Intel chipsets, including the 845MP, 845MZ, 852PM, 852GME and 855PM (when used with discrete graphics)."
so 4 different companies somehow magically made the exact same flaw that only appears on the mentioned Intel chipsets?
did YOU even read it?
It is possible if they simply copied a reference design that one of the companies put out.
And what in the article says its an Intel problem? The text of the article clearly states that the problem is with the memory modules and not the chipsets.
I should also add that the 845xM and 855xM series chipsets are different designs by different design teams (845 is a desktop division product, while 855 is a notebook division, Intel Israel design).
I do not see why people are jumping to conclusions that this is an intel problem. If there were serious memory problems with common modules, the Dell support forums would be alight with infuriated end users.
And what in the article says its an Intel problem? The text of the article clearly states that the problem is with the memory modules and not the chipsets.
I should also add that the 845xM and 855xM series chipsets are different designs by different design teams (845 is a desktop division product, while 855 is a notebook division, Intel Israel design).
I do not see why people are jumping to conclusions that this is an intel problem. If there were serious memory problems with common modules, the Dell support forums would be alight with infuriated end users.
It seems that it would be close to impossible that multiple memory mfg would have a inherent design flaw, which does not show up in ANY other chipsets besides the aforementioned Intel mobile Pentium designs but as was menitoned it could have been a reference design problem?
But, I would argue that even if the flaw is with the memory modules, since these modules are the de-facto standard memory module, it is Intels responsibility to make their chipset compatible with the standard memory designs and NOT vice-versa.
Although, I do applaud these companies for recalling these poorly designed chipsets.
VirtualLarry
No Lifer
Via had designed P4 chipsets, without a license to the US patents covering the P4 CPU bus. So the result of that was that Via could sell those chipsets in other areas of the world, but could not sell/import those chips into the US. Via attempted to get around that issue by purchasing another company that did have a P4 bus license, so they got a sort of indirect license that way. (S3? I don't remember.)
I think that they sued each other (Intel and Via), and finally settled out-of-court with Via paying for a P4 bus license like they had originally wanted to, because Intel was taking some heat from a FTC anti-competitive practices investigation. Something like that anyways.
Yes, the 686B IDE-corruption/dataloss error was pretty severe. (Worthy of a chipset recall IMHO. In fact, look at what happened to those notebook mfgs that shipped systems with an old-revision NEC floppy controller that had some obscure data-corruption error, and they got hit with a major class-action lawsuit. Yet Via gets off scott-free, for releasing multiple generations and revisions of products, all with similar/same errors, and much more likely to be triggered by "real-world" scenarios. Go figure.)
But what I was talking about was a performance-related issue with their PCI bus implementation, due to lack of bus-park functionality which is still largely true today. (At least up until their 8235 which I am currently using, unsure if the issue was fixed in the 8237 or not. Not enough reported data to tell.)
By changing the PCI latency settings, the device in question would be allowed to use the bus for longer bursts, but it would still be periodically forced off of the bus, even if it was the only device using the bus. Intel's design allowed devices to basically hog the bus (as long as no other device requested use of it), and therefore obtain bandwidth much closer to the theoretical maximum.
KristopherKubicki
Golden Member
You don't see AMD recalling all their Opterons...
On the other hand... AMD's mistake isnt actually something that has been replicated in the wild yet. I think this was in their press release.
Kristopher
On the other hand... AMD's mistake isnt actually something that has been replicated in the wild yet. I think this was in their press release.
Kristopher
zerodeefex
Senior member
hahahahahaha, my gf's dad tried to tell me (not knowing my mom worked for AMD for over 15 years), that "AMD is going to break. We have a lot more testing and quality assurance. They have serious quality issues. We never have any problems after it leaves the factory."
Edit: Forgot to mention he works at Intel.
Schadenfroh
Elite Member
sorry man, i could never date a girl whose father worked for intel
One way that it could be a problem with specific chipsets and specific memory modules is if the traces on the board layout on the PCB used timing information that relied on the chipset and modules meeting a certain specification target for timing and then this wasn't actually the case in reality.
I have no idea what the problem is aside from reading the link, but it seems to me that the way it was phrased they are laying blame on the memory modules, but the only way that otherwise good memory modules won't work in a specific case is if the module somehow doesn't entirely meet the specification - or if there's a misunderstanding of the specification. There are three parts to the system here: the memory, the chipset and the PCB board. My wild guess is that HP's design relied on the memory responding in a certain way to a C3 power management request according to a specification and that certain modules are not doing what HP's designers expected.
In the past things like this have happened before; who doesn't remember the initial teething problems of PC100 memory where only certain memory would work in certain motherboards, or that certain motherboard vendors would recommend only certain memory be used?
Please note that although I work for Intel, I have absolutely no insider knowledge of this problem and am merely speculating a possibility that I didn't see anyone else mention based on the information that is in the press release.
I have no idea what the problem is aside from reading the link, but it seems to me that the way it was phrased they are laying blame on the memory modules, but the only way that otherwise good memory modules won't work in a specific case is if the module somehow doesn't entirely meet the specification - or if there's a misunderstanding of the specification. There are three parts to the system here: the memory, the chipset and the PCB board. My wild guess is that HP's design relied on the memory responding in a certain way to a C3 power management request according to a specification and that certain modules are not doing what HP's designers expected.
In the past things like this have happened before; who doesn't remember the initial teething problems of PC100 memory where only certain memory would work in certain motherboards, or that certain motherboard vendors would recommend only certain memory be used?
Please note that although I work for Intel, I have absolutely no insider knowledge of this problem and am merely speculating a possibility that I didn't see anyone else mention based on the information that is in the press release.
mrSHEiK124
Lifer
poor guy, no one replied to ya. well, i don't know much about intel's chipsets (too confusing to keep track of), but it said it was an issue with Mobile P4s, which are completely different, much more inferior chips to 😀 the centrinos (Pentium M, based off part of the PIII technology, not the gay NetBurst which was designed for the sole purpose of achieving insane clock rates on bandwidth starved CPUs)
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 25K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 24K
-
-
