Mushkin DDR2 1150 won't even do 1066?

lash_larue

Limp Gawd
Joined
Nov 26, 2006
Messages
149
i got 2 gigs of corsair dominator ddr2 800 that i bought off a forum member here that will run damn near any setting i put on it but the mushkin ram constantly errors out.

last night i swapped out my dominator and put in the mushkin and made a pass or two with memtest with no errors. keep in mind whatever setting the bios was running, my ddr2 800 dominator was running it with no memtest errors (have made many passes with it). earlier tonight i changed a bios setting and memtest was full of errors. so i change the settings back, same problem. bios has presets where you save your settings, and i had mine saved so i reloaded and STILL have errors with the mushkin. i have upped and lowered the voltage from 2.1 to 2.5 to no avail.

factory voltage is 2.3, so i set that and set all other processor/ram settings to auto/default (no overclocking), still errored out with memtest. let the board use auto voltage of 1.8 with all factory settings and it tested fine, but it was running at ddr2 800. so i have a setting in my bios fsb (Bsel) to Northbridge Latch, set it to FSB 1333, and set my DRAM frequency to DDR2 1066. no other overclocking at these settings, cpu frequency is at 333. it still errored out with no processor overclocking, factory 2.3v, and the 1066mhz setting.

if this mushkin is ddr2-1150 shouldn't it be able to handle 1066? that was the point of buying it. i've got 130 bucks tied up in it and it doesn't perform as well as the corsair ddr2 800 i bought for 40 bucks.

if the info is needed all ram is 1gb sticks and i am not mixing it together, i'm running it paired. the mushkin timing is 5-5-4-12 and the corsair timing is 4-4-4-12 but it's pc2-6400.
 
they said it could do this speed.

Your's cant do this speed.

Sounds like RMA to me...
 
sorry i had posted it and edited out that line and forgot it, it's a biostar tpower i45.

i put my 800mhz dominator back in and it runs memtest fine at ddr2-1039 setting, set to read from the spd chip. i've been around the block a time or two with this overclocking stuff but i'm pretty new to overclocking core2duo's, so am i right in thinking that according to that setting my ddr2-800 rated ram is running at 1039mhz? i realize there's all that double and quad pumped hooplah, but nonetheless is that pretty much correct?

i agree with the guy that says RMA but what pisses me off is that it seems the majority of this ram i buy never runs like it's supposed to. i got 2 gigs of corsair xms ddr400 that i have in my 939 work rig that will barely do what it's rated to do. the first 2 i bought with the computer were bad all along and i didn't find out until months later. after all who buys brand new ram and expects it to have errors, thought they had quality control? bought 2 more 2gb kits and one stick in each one was bad. so i had to go through a total of 6 sticks to get 2 that worked.

8 or so years ago i bought a 128meg mushkin stick, pc150 and it wouldn't go past 133. overclocking was still a pretty new hobby then so i thought big deal. i would've thought they had the bugs ironed out by now... /end whiny bitch mode
 
well mushkin wasn't very helpful. i quote captain obvious:

"Sounds like you can't use that voltage without it errors, try finding the point in which voltage start causing it to error or setup an rma."

found that to be a pretty dumb ass response considering the point it starts to give errors is at the factory specified 2.3v... i'll rma it alright, right back to newegg.
 
a fugly 3.3V rail from an ugly PSU can give the memory trouble.

and yeah, to figure out all the bus speeds I think its best to look at it from the north bridges perspective. you've got a bus between the NB and the processor which you have total control over, everything else has to be a spin off of that speed. So if you've got your FSB at 400MHz the memory speed has to be a spinoff of that.

I'm betting you get the ratio whooplah but i suspect some who are reading this thread don't so I'll go over it, at stock speeds:
FSB core: 266MHz, FSB effective: 1066MHz, Ratio 1:1, RAM actual: 266MHz, RAM "DDR" (effective) : 533MHz.
FSB core: 266MHz, FSB effective: 1066MHz, Ratio 4:5, RAM actual: 333MHz, RAM "DDR" (effective) : 667MHz
FSB core: 266MHz, FSB effective: 1066MHz, Ratio 2:3, RAM actual: 400MHz, RAM "DDR" (effective) : 800MHz

to get your ram to 1066 using the minimum bus speed needed you would use:
FSB core: 356MHz, FSB effective: 1424 MHz, Ratio 2:3, RAM actual: 533MHz, RAM "DDR" (effective) : 1066MHz
 
actually i meant i understand the speed terminology, to be honest the ratios are somewhat confusing but if i'm using the DRAM frequency setting to set the speed doesn't that take care of worrying about the dividers?

i like to think i have a good psu, it's an antec quattro 1000w. the mobo shows it putting out 11.88 constantly for the 12v reading but i tested it at the ATX connector with a volt meter while running crysis benchmarks and it never dipped below 12.28. i dont know if my board has issues or is crap, or if the winbond chip that reads that info is just crappy. the bios shows the 3.3v at slightly over or slightly under, within range according to the atx spec sheet someone posted in another thread for me.

i'm more than likely gonna return the ram to newegg and i'm thinking about returning the motherboard and maybe trying another.
 
Are you running the timings by SPD, or setting them manually? Have you tried each stick by itself to see if only one is causing an issue?
 
@fusion - setting the timings manually or by SPD didn't help either way. I have not tested each stick individually because if they can't work together in my board its the same as broke to me. If there weren't restocking fees it would be a non issue and i'd be glad to tinker but since returning it costs me money I see it as an inconvenience. As far as RMA'ing it to Mushkin I'll pass on that, the half baked answers i got on their forum has told me all I need to know.

@moelarryncurly - hehe at the name, and i'll read through that link. thanks for the info.
 
Any company that sells memory modules with UTT chips on it doesn't really care about quality.

UTT chips:http://www.legitreviews.com/article.php?aid=218

wow thats pretty crappy - and to think the moderator on their forum told me they hand test every stick before it leaves.

be my luck i could buy 3 different brands to replace what i'm sending back and probably have at least 1 bad stick in each one. anyone got any recommendations other than corsair? i would like to get a 4gb (2 x 1) kit that will overclock fairly decent but with that size I don't know if thats a realistic goal or not.
 
wow thats pretty crappy - and to think the moderator on their forum told me they hand test every stick before it leaves.

be my luck i could buy 3 different brands to replace what i'm sending back and probably have at least 1 bad stick in each one. anyone got any recommendations other than corsair? i would like to get a 4gb (2 x 1) kit that will overclock fairly decent but with that size I don't know if thats a realistic goal or not.

That moderator might still be right. Micron, Nanya and one devision of infineon don't manufacture the sticks of ram, they simply manufacture the silicon chips. Companies like Mushkin, Patriot, OCZ, and the subsiderary company's of Infineon and Crucial take these chips, strap them to PCBs, test them to variing degrees of stress, ship them, and provide support (or lack there of) to the consumer. So Mushkin might still be "hand testing" their sticks. But to that point, what good is "hand testing". Why is hand testing any better than machine testing? Its like how KFC always advertises "hand breaded" chicken, what, your machine is broken?
 
Well I figured after the 3rd color he would get the idea. lol
However, the redline is really tops.
Though Mushkins really OC well in general
 
As far as RMA'ing it to Mushkin I'll pass on that, the half baked answers i got on their forum has told me all I need to know.

I've looked at our forums, and I can't find a post regarding the XP2-9200 kit, could you link your post to me?

o Mushkin might still be "hand testing" their sticks. But to that point, what good is "hand testing". Why is hand testing any better than machine testing? Its like how KFC always advertises "hand breaded" chicken, what, your machine is broken?

There's a big difference between machine testing and using actual motherboards and software to do the testing. I know some companies are moving more to the DIMM testers that are out there to up their throughput, but nothing will beat actual platforms that people will be using with their RAM.
 
A fancy DIMM tester, like the $$$ ones that Samsung and Micron use, can check the quality of each signal and indicate how well they meet specs, accurately enough that DIMMs can be sorted for speed with them. OTOH a mobo is just a pass/fail tester for DIMMs because it can't resolve signals beyond a half clock cycle for DDR2 and DDR3.

I've had to test some DIMMs in as many as four different types of mobos before errors cropped up. I don't know if Mushkin has enough time to do that.
 
Back
Top