Archived

This topic is now archived and is closed to further replies.

Sesquashtoo

Seems high-performance Mem Modules are failing....

Recommended Posts

...enough to have some manufacturers suspend sales of their product line!The Mushkin Redline Ascents have been deactivated at Newegg, AND..every one of them (different timings) are no longer for sale RIGHT FROM MUSHKIN!Not rocket science...the memory chips at aggressive timings are not holding up. Glad that I stuck for now with my stock 1066's oc' to 1367 MHz. I had been following the page waiting for a good price...and after seeing these were deactivated, went to www.mushkin.com, and found that they ALL (Ascent line-up) are not for sale. Not out of stock---not for sale. Some FROSTBITE's still are. Interesting development, being that some other AVSIM users were having module failures (also other than Mushkin...Corsair, etc) with aggressive-timing products. It will be interesting to follow this development over the next few months of the new year![/url]http://www.newegg.com/Product/Product.aspx...N82E16820226050

Share this post


Link to post
Share on other sites
Help AVSIM continue to serve you!
Please donate today!

I've been researching DDR3 memory for the past few days and recalled seeing a comment that high performance Elpida Memory chips provided to several companies including Corsair, OCZ, and Kingston were causing serious problems. In a July 22, 2009 press announcement Corsair noted it had "re-introduced the popular Dominator GT line of memory modules. These modules had been removed from the market due to unacceptable failure levels of the Elpida "Hyper" components with which the modules are constructed." Sounds like it's now Mushkins turn.

Share this post


Link to post
Share on other sites

I thought those were only the DDR3 running at 2000MHz ???Hmmm, those mushkin kits (were) some of the best... too bad....

Share this post


Link to post
Share on other sites

You do realize there is a clock on the motherboard so you can NEVER achieve 2000mhz performance...or even 1600 for that matter...gate is open information passes through gate is closed, and repeat... unless some of you have some your own theories on synchronization :) If you think information is just passed through and it's written at 2000mhz speed, you really need to take a course on electronics.

Share this post


Link to post
Share on other sites
You do realize there is a clock on the motherboard so you can NEVER achieve 2000mhz performance...or even 1600 for that matter...gate is open information passes through gate is closed, and repeat... unless some of you have some your own theories on synchronization :) If you think information is just passed through and it's written at 2000mhz speed, you really need to take a course on electronics.
It appears that you do not understand what a 2 GHz memory clock speed really is. Maybe you need to go take one of those courses, too.CheerioBob ScottColonel, USAF (ret)ATP IMEL Gulfstream II-III-IV-VColorado Springs, CO

Share this post


Link to post
Share on other sites

While were on the subject of memory here, and with all other things being equal (the memory appropriately matched / balanced and setup in a good CPU speed-wise for FSX), what sort of performance difference could be expected going from a CAS9 to say a CAS7 timing in use with FSX?

Share this post


Link to post
Share on other sites
It appears that you do not understand what a 2 GHz memory clock speed really is. Maybe you need to go take one of those courses, too.CheerioBob ScottColonel, USAF (ret)ATP IMEL Gulfstream II-III-IV-VColorado Springs, CO
You have no idea what your talking about do you? memory clock speeds are derived from the system clock... they are not separate entities. You ever heard of clock jitter ?

Share this post


Link to post
Share on other sites
If you think information is just passed through and it's written at 2000mhz speed, you really need to take a course on electronics.
No sorry, I don't do electronics hard core, just air traffic control.AFAIK the 2000MHz RAM was mainly for the hard core overclocking club...

Share this post


Link to post
Share on other sites
You have no idea what your talking about do you? memory clock speeds are derived from the system clock... they are not separate entities. You ever heard of clock jitter ?
So what?? The base clock is a system reference clock that is then multiplied or divided as needed by other system controllers. The independence of the clocks is irrelevant. Now you suggest that "2000 MHz performance" (whatever that means) is not possible because the BCLK is on the motherboard? Wrong. I guarantee you that a 2 GHz mem clock speed is trivially attainable on an unlocked CPU even at the default 133.33 MHz BCLK by setting the memory clock multiplier appropriately.Yes, I know what clock jitter is...and how to quantify it with a spectrum analyzer and a few measurements of the phase noise envelope. And once again, a big "so what?"RegardsBob ScottColonel, USAF (ret)ATP IMEL Gulfstream II-III-IV-VColorado Springs, CO

Share this post


Link to post
Share on other sites

None of what you said makes sense... obviously because you're trying to quote wiki... if you don't even know the gate limitations of ram then I doubt you understand anything. We're talking about jitter in terms of impulse trains.. NOTHING ELSE... spectrum/phase/noise have nothing to do with the topic on hand. Unless you think a single point is a "spectrum"

Share this post


Link to post
Share on other sites

This is the same fallacy that existed ever since system clock, CPU clock, and memory clocks all diverged at the beginning of last decade. Just because you have differential clocks doesn't mean they're unmatched. Interfaces have widths as well as speeds. You need to multiply the two to derive a performance metric.

Share this post


Link to post
Share on other sites
None of what you said makes sense... obviously because you're trying to quote wiki... if you don't even know the gate limitations of ram then I doubt you understand anything. We're talking about jitter in terms of impulse trains.. NOTHING ELSE... spectrum/phase/noise have nothing to do with the topic on hand. Unless you think a single point is a "spectrum"
OH NO!!!! Not another Mr Know it all.... Here we go again :( What on this earth does all that mumbo jumbo have to do with enjoying flightsim. It really doesnt belong in a flightsim forum. Other than letting other users know that there are bad modules out there and to be cautious if considering buying them which is exactly what the op did, all this indepth back and forth technical crap that means absolutely nothing to us should be left to more technical forums where you can argue back and forth at your hearts content.

Share this post


Link to post
Share on other sites

these aren't know it all concepts... synchronization concepts are as basic as they get. The real problem is idiots who buy this stuff, then claim there is a difference that's worth spending twice as much. Honestly the most genius comment comes from the guy who's overclocking his ram, instead of underclocking it.

Share this post


Link to post
Share on other sites
these aren't know it all concepts... synchronization concepts are as basic as they get. The real problem is idiots who buy this stuff, then claim there is a difference that's worth spending twice as much. Honestly the most genius comment comes from the guy who's overclocking his ram, instead of underclocking it.
The differences between speeds were/are measurable, and therefore worth the relatively small extra cost (to me). Huge difference? No. But noticeable nevertheless.That said, I think I'm going to take NickN's lead and simply refuse to argue with name-calling 10th graders that think they're quantum physicists. 'nuf said.RegardsBob ScottColonel, USAF (ret)ATP IMEL Gulfstream II-III-IV-VColorado Springs, CO

Share this post


Link to post
Share on other sites
these aren't know it all concepts... synchronization concepts are as basic as they get. The real problem is idiots who buy this stuff, then claim there is a difference that's worth spending twice as much. Honestly the most genius comment comes from the guy who's overclocking his ram, instead of underclocking it.
Granted... but telling people that they need to take a course in electronics, talking about garden gates, clock wobblies and frieght trains :( I think I can speak on behalf of alot of us here that we definately appreciate and definately gain more valuable knowledge with hardware decisions by well informed posts that keep it simple. Tech gurus going at it with absolutely no other purpose other than to prove the other guy wrong can be quite entertaining but really how many times does the OP walk away with a satisfactory answer.

Share this post


Link to post
Share on other sites

I believe I made the same argument when these came on the market. You guys listened to the so called experts and some people are now stuck with very expensive junk. Maybe if you stopped listening to these people and their so called informed posts than this thread wouldn't have been necessary??? Bottom line the real tech guys are going to give you a YES/NO answer. Not tell you what you want to hear, so they can justify their own or your stupid purchases. If an explanation sounds too simple.. question it. But don't confuse the direct YES/NO answers validity because they lack explanation, bottom line is the real answer is far too long for a forum posts. Tech guys are not going to sit there and tell you exactly how to build parallel NAND circuits just to explain how RAM works.

Share this post


Link to post
Share on other sites

I still have both my original 6-5-6-18 and the 6-7-6-18 Mushkin sticks and they work just fine.. I had one set I bought as high-clock sticks from Corsair and another from OCZ that had a bad stick in them and were RMA'd And I had one set (extra on the 6-7-6) from Mushkin I had to RMA for one stick This issue was seen by all the memory companies and not just Mushkin.. and it had nothing to do with the timing of the modules either.. Yes Bob, I agree which is why I do not visit here very often any more other than to respond to threads I may have posted in the past. If the 'know-it-alls" actually knew the backgrounds involved they were challenging they wouldn't open their smart mouths very often and prove exactly how old they are or how much they really do not know.

Share this post


Link to post
Share on other sites

This from a guy who's optimizations reduce frame rates by half...

Share this post


Link to post
Share on other sites

BionicCrab - This thread is filled with provacotive and unnecessary posts from you. There's no need to be so aggresive amongst a crowd of folks just trying to give each other advise and provide a welcoming community of sharing ideas and suggestions. Tone it down ASAP, if not sooner.

Share this post


Link to post
Share on other sites

I just find it odd that anyone would actually debate whether or not SDRAM(Synchronous Dynamic RAM) timing was controlled by the system clock or not :) I mean if it seems like I'm trolling fine... but at the very least the only two people who are disagreeing are the ones who claim to have an electrical engineering background when it's obvious they don't. What's worse they use these so called "qualifications" as a justification for their wild theories on other things, including "shader based graphics cards", when in reality the entire pipeline is reprogrammable, or "pre-rendered graphics" being an actual real-time rendering method. Now, a certain someone is claiming he never actually recommended a particular defective product line, and in fact uses a lower CAS delay than what he previously recommended and argued vehemently was equivalent. To me it's plain conjecture not expertise.

Share this post


Link to post
Share on other sites

Timings are not a measurement of performance. Latency however, is. Latency of course is derived by dividing minimum access time (in clock cycles) by clock cycles per second. This usually provides a latency in the 1's or 10's of ns (nanosecond) range, when discussing latency of modern consumer-class DRAM devices. I won't comment on others directly. No need to get personal with anyone. I agree that sometimes terminology is mis-used and abused around these parts, however.

Share this post


Link to post
Share on other sites

Just want to re-amplify for those who have access to (and are still considering buying) the part number listed at the top of the thread (998691).The high failure rates were associated with Elpida

Share this post


Link to post
Share on other sites

I just lost one of my Mushkin Redline modules part number 998692 6GB (3x2GB) DDR3 PC3-12800 6-7-6-18 Redline (Triple Kit). I have been running this memory at the manufacturers timings and voltage. There must have been problems across the board with the redline products as they were EOL'd pretty quick. I also notice that no manufacturer at present sells a triple channel kit at these timings including Mushkin. I purchased mine in September less than 4 months ago and one module is completely dead. I can tell you that there is a very noticeable difference running FSX with the memroy running in dual channel mode instead of tri-channel mode. Bummer.I already ordered some Corsair Modules and am awaiting a response from Mushkin support.

Share this post


Link to post
Share on other sites