backtop

Print E-mail del.icio.us 22 comment(s) - last by Zoomer.. on May 19 at 6:35 AM
Enhanced Performance Profiles are here, but JEDEC does not condone them

Yesterday we ran an article about a new memory standard, dubbed Enhanced Performance Profiles.  The new NVIDIA backed memory proposal uses areas of the JEDEC SPD data and command areas for specific hooks that allow upcoming nForce motherboards to overclock automatically and allow for fine tune control of memory voltages and frequencies.

NVIDIA cast the first stone at JEDEC, the standards committee for solid state memory devices, claiming existing JEDEC specifications are missing critical specifications for overclocking. NVIDIA's EPP memory proposal claims EPP will bring an open standard to memory manufacturers and that the platform has been submitted to JEDEC for approval.  However, the Chairman of the JEDEC SPD group issued the following statement to memory manufacturers earlier today:
    This week an industry announcement was made for a “New Open Standard Memory Specification” called “Enhanced Performance Profiles” or EPP.  This proposal suggests that it can “allow memory manufacturers to integrate additional module performance data” by redefining the manufacturer’s data region, bytes 99-127.
     
    As chairman of the JEDEC SPD Task Group, I feel it is my responsibility to inform the committee that this EPP specification was not developed in my task group.  In fact, we have not seen a first showing on the topic.

    EPP is not a JEDEC endorsed specification and should not be misconstrued in the industry as such.  I do encourage the sponsors of the concept, or any other companies wishing to properly document higher DDR2 speed grades, to develop a truly open specification by submitting a proposal to the JEDEC committee and to the SPD Task Group and following the well established guidelines by which open standards are developed.
Hopefully JEDEC and NVIDIA can resolve and approve such a standard, though JEDEC has been extremely hesitant to approve technologies that run out of band.  Considering that NVIDIA's EPP purposely runs out of specification, it may take a bit of negotiating on both sides to come to a middle ground on this issue.  Previous standards proposed to JEDEC have sometimes taken years to get final approval, which means NVIDIA's Tritium program may already be facing hostility before it gets out the door.

Update 05/16/2006: NVIDIA public relations has contacted DailyTech claiming EPP has not been submitted to JEDEC.  Instead, NVIDIA's Bryan Del Rizzo says "We told editors who were briefed beforehand that we would be submitting to JEDEC for possible ratification at the earliest possible opportunity."

Comments     Threshold


This article is over a month old, voting and posting comments is disabled
Not truly open?
By appledelhi on 5/16/2006 9:56:04 PM , Rating: 2
This reminds me a little of the Rambus/Jedec situation.




RE: Not truly open?
By PrinceGaz on 5/16/2006 10:21:05 PM , Rating: 1
What nVidia are promoting with their "overclocking data for supported nVidia chipsets" is proprietary nonsence that has no place in an industry-standard that is essential to the reliable operation of the vast majority of computers around the world. The whole point of a standard is that it can be used with any compatible component; as soon as you start tagging on extras for this and that the whole idea falls apart as everyone and their dog is going to add their own standard.

Imagine if nVidia gets away with having their data added to the SPD. You can be sure ATI will also want to have data added to the SPD in modules as well for automatic optimum performance. Intel will want data included for their future high-performance chipsets (like today's 975X) if they're to compete.

Before long we'd be in the crazy situation where instead of the SPD giving a set of timings the memory should work at with any validated platform, individual memory modules include a whole bunch of different timings for individual chipsets from several manufacturers. That means extra work for memory-module manufacturers to validate modules on every platform which will be added to the retail cost that we all pay, all for a tiny fraction of a percent improvement in overall system performance. The difference between 4-4-4 and 5-5-5 with DDR2 is under one percent, so it goes without saying the timing tweaks on the less important settings nVidia have added will be lucky to add more than a small fractional percent gain in performance. Gains that any overclocker worth his salt could have made with manual tuning, along with a bit more besides (as the tweaked timings will be "safe" overclocks).

SLI memory??? What does SLI stand for this time? It can't be scan-line interleave or scalable-link interface. Maybe it is Scam Lamer Idiots as only fools would think this ploy from nVidia is anything short of a ploy to sell overpriced products that perform no better than any others.


RE: Not truly open?
By theapparition on 5/17/2006 7:38:53 AM , Rating: 2
Obviously, you don't work in the industry or have much knowledge of how these things work. Manufacturers have for decades implemented proprietary extensions and then submitted them for approval to a standards organization. Almost all hardware on the modern computer has been developed and marketed by a single manufacturer and then accepted as the standard. Nvidia has publicly stated they intend to submit these extensions and keep them open.
One problem with oversight committees is that they tend to move very slowly. Get a room full of competing companies, each with its own agenda, and the agreement of a certain specification can take years. I'm sure the average computing community prefers to not wait several years for the next advancement. Imagine getting a standards organization together for CPU's. How long do you think it would take Intel, AMD, Via to agree on a completely compatible standard, where chips were as interchangeable as memory. Seems ridiculous to even entertain the thought!
I have no issue with a manufacturer using unused portions of the SPD hook to implement a proprietary solution.....so long as they keep that solution open, and try to get it a standard. That's innovation in my book! Take CompactPCI for instance. Over half of pins on a 6U backplane are unused. Why are they unused? To allow future revisions and extensions. Many companies then developed extensions that were submitted to PICMG such as VMD over CPCI, Telephony over CPCI, etc, where some of those pins were now defined.
With that said, I personally HATE closed door specifications and proprietary solutions. So I understand your initial resentment. However, the difference is that any memory module will work in these motherboards. You just won't get the performance increase. If you bought one of these SLI memory modules, it would still work in your non-Tritium motherboard (Don't know why someone would pay the extra, but for sake of argument). So there is a difference with truly proprietary hardware and extensions that increase performance or compatibility.
Down with closed door/proprietary hardware. I'll never buy a Sony product for that very reason. Beta, MiniDisc, MemoryStick, UMD...........could go on forever. Also, that's why I hate Apple. If they ever open up Fairplay, I'll get an iPod. Not one moment before that. And we all know that's a snowblower's chance in hell.


RE: Not truly open?
By Trisped on 5/17/2006 1:56:02 PM , Rating: 2
quote:
Many companies then developed extensions that were submitted to PICMG such as VMD over CPCI, Telephony over CPCI, etc, where some of those pins were now defined.
But that isn't what NVIDIA did. They developed the extensions, told everyone about them and said that they were going to use them, THEN told JEDEC that they would like them considered. Like the article says, there is a process, follow it.
(I agree with your dislike of the proprietary. Low value, high cost.)


RE: Not truly open?
By Gentleman on 5/16/2006 11:35:37 PM , Rating: 2
Why take the fun out of overclocking? Part of the adventure is to discover how far your components can go by playing around with the settings.


RE: Not truly open?
By clementlim on 5/17/2006 2:20:57 AM , Rating: 2
Hey, not everyone is as adventurous and smart as you, you know. Only technocrats, like people here at DT, know what and how to overclock without destroying the computer. My Dad don't even dare to open the casing of the computer. Some people want the computer to be faster but don't dare to overclock it themselves. I guess what nVidia is doing is great, conceptually that is. Price is another matter.


RE: Not truly open?
By Stele on 5/17/2006 4:00:12 AM , Rating: 3
imho, that's like saying some people want to drive their cars at 200MPH but don't know/want to learn how to handle a car at such speeds. If you want to do something, you ought to learn it right and proper from the basics. If certain people don't want to do so, they shouldn't be messing around in things they don't know enough about in the first place.


RE: Not truly open?
By Zoomer on 5/19/2006 6:35:29 AM , Rating: 2
If you are afraid of overclocking, get the FX60 instead of the Opteron 170. It's guaranteed to be faster.

Same for all of these. Such rubbish like automatic overclocking is stupid. If they are validated by the manufacturer, technically you aren't overclocking as the product has been designed, or at least validated, to work at that speed.


RE: Not truly open?
By Gentleman on 5/16/2006 11:39:53 PM , Rating: 2
Why take the fun out of overclocking? Part of the adventure is to discover how far your components can go by playing around with the settings.


JEDEC approval?
By bob661 on 5/16/2006 10:14:41 PM , Rating: 2
quote:
which means NVIDIA's Tritium program may already be facing hostility before it gets out the door.
If JEDEC doesn't approve, Nvidia and company will just do it anyways. Nvidia can always keep the standard open to other manufacturers.




RE: JEDEC approval?
By PrinceGaz on 5/16/2006 10:25:04 PM , Rating: 4
And you think it's a good thing if nVidia ignores JEDEC and encourages memory-manufacturers to produce non-standard modules all for the sake of a tiny performance-boost on certain supported chipsets? You are out of touch with the real world.


RE: JEDEC approval?
By bob661 on 5/16/2006 10:32:11 PM , Rating: 2
quote:
You are out of touch with the real world.
Ok.


RE: JEDEC approval?
By Trisped on 5/17/2006 1:47:54 PM , Rating: 2
You are forgetting the other problem. The standard is JEDEC's, not NVIDIA's. NVIDIA has basically invaded it and encouraged manufactures to support theirs. When JEDEC release their next up grade people will have to worry about compatibility. Will the NVIDIA changed products be compatible with the new JEDEC? Will JEDEC's work without errors with NVIDIA? JEDEC probably sees it as a slap in the face. All that work to make things "just work" so the average consumer won't have problems using it, and now NVIDIA is trying to black-mail them to get their way.


its not non-standard
By rika13 on 5/17/2006 6:43:35 AM , Rating: 2
the normal jedec spd info is still there and accessible by everyone

what nvidia is doing is adding more information in an unused area, nothing abnormal here

jedec is just annoyed that they dont have control over it




RE: its not non-standard
By DigitalFreak on 5/17/2006 7:11:02 AM , Rating: 2
Yeah, they are plenty pissed. </sarcasm>

What happens when the JEDEC decides to add it's own, standard, data to the area that Nvidia is using?


RE: its not non-standard
By Master Kenobi (blog) on 5/17/2006 9:26:32 AM , Rating: 3
Then you get battle lines to form up overnight as everyone sides with JEDEC or nVidia, and we get an industry split.


RE: its not non-standard
By Trisped on 5/17/2006 1:39:56 PM , Rating: 2
Actually it is used, just not in currently available products. If you read the previous news posting you will see, "By adding commands in the SPD area that currently have been reserved for future SPD expansion."

JEDEC had left a spot for extra commands. This is common in the computer industry because needs change so much you need them to be upgrade able. They have probably been working on things that need to be stuck in there for the next JEDEC update.

I don't know about you, but any time I am working on a project and someone comes in and say, "I don't care what you are doing or why, I am going to do my own thing and spend enough money to make all your work seem worthless or problematic."


Simpsonesque
By covertbit on 5/17/2006 9:30:08 AM , Rating: 2
[Nelson Voice] HA HA [/Nelson Voice]

Why must nVidia throw around SLI like it is going to sell more things in areas other than video? Next thing u will see, "SLI teh FuX0rS out of your Gigabit network interfaces to handle our proprietary dual-gigabit speeds!". And in small lettering, you must buy nVidia switches, cables, software, routers, etc etc for it to run correctly. I dunno, this kid of crap doesn't help the industry, it makes it more cumbersome. Give me a stable memory controller and I will overclock the fuck out of it, I dont need a computer doing it for me!




RE: Simpsonesque
By Master Kenobi (blog) on 5/17/2006 9:39:00 AM , Rating: 2
Well were getting into why nVidia can sell. SLI is their latest gimmick. You can walk into a BestBuy and say hey you got Crossfire? And they will look at you funny. Tell them you want SLI, and they are all over you like stink on crap. It's a coined phrase that people are attributing to "High Performance" when its really just 2 video cards married up. Now nVidia is promoting the SLI not "How it is", but "How people look at it", which is "high performance", so less informed people (The vast majority of the mainstream sadly) will simply see SLI Mobo, HD, Memory, etc.. etc... and will think wow thats a whole lot of High Performance stuff, this machine must kick ass! So they can get away with attaching a higher pricetag.

nVidia has taken a page from the Sony marketing handbook here. Ride the hype, the hype says you are a "Superior Quality" company, and that everything you make is "Superior Quality", true or not is irrelevenat, its what the Hype is. So they take the SLI name, attach it to everything, and raise the price by 10-15% per item, and they can sell it because it says SLI, and nVidia will say, SLI is a superior quality and superior performing to all other types of hardware. Sony has done this for years, its why they can sell highly priced stuff thats par or sub-par and get away with it. nVidia is doing this now. Welcome to marketing 101 folks, enjoy the stay.


RE: Simpsonesque
By Trisped on 5/17/2006 1:30:17 PM , Rating: 2
Don't forget the Sony proprietary. It is the same with SLI.


By Trisped on 5/17/2006 1:23:41 PM , Rating: 2
quote:
I do encourage the sponsors of the concept, or any other companies wishing to properly document higher DDR2 speed grades, to develop a truly open specification by submitting a proposal to the JEDEC committee and to the SPD Task Group and following the well established guidelines by which open standards are developed.
Sounds like JEDEC is upset that NVIDIA is not following proper submission procedure while trying to get a proprietary format made standard.

Are they really trying to open the standard, or do they want to charge everyone who does not play by their rules?




"When an individual makes a copy of a song for himself, I suppose we can say he stole a song." -- Sony BMG attorney Jennifer Pariser











botimage
Copyright 2009 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki