Mushkin Redline XP4000: Winbond with Voltage Be Damned
by Wesley Fink on May 16, 2005 12:05 AM EST- Posted in
- Memory
Test Results: Mushkin Redline XP4000
The top row of Performance results at 2.6V is included as a performance baseline. Like OCZ VX, Mushkin Redline cannot achieve 2-2-2 timings at stock voltage at DDR400. However, it can easily reach 2-2-2 performance levels at DDR400 with more voltage.Mushkin Redline XP4000 (DDR500) - 2x512Mb Double-Bank | |||||||
CPU Ratio at 2.4GHz | Memory Speed |
Memory Timings & Voltage |
Quake3 fps |
Sandra UNBuffered | Sandra Standard Buffered |
Super PI 2M places (time in sec) |
Wolfenstein - Radar - Enemy Territory fps |
12x200 | 400 DDR (Stock Voltage) |
2-3-2-6 2.6V 1T (Stock Voltage) |
565.4 | INT 2719 FLT 2866 |
INT 6111 FLT 6056 |
81 | 119.9 |
12x200 | 400 DDR | 2-2-2-6 3.0V 1T |
572.5 | INT 2745 FLT 2899 |
INT 6127 FLT 6071 |
81 | 119.8 |
11x218 | 436 DDR | 2-2-2-6 3.1V 1T |
578.0 | INT 2912 FLT 3079 |
INT 6512 FLT 6435 |
80 | 120.2 |
10x240 | 480 DDR | 2-2-2-6 3.2V 1T |
591.7 | INT 3037 FLT 3293 |
INT 6775 FLT 6700 |
79 | 122.7 |
9x267 | 533 DDR | 2-2-2-6 3.4V 1T |
599.1 | INT 3308 FLT 3521 |
INT 7125 FLT 7032 |
78 | 124.0 |
9x272 (2.45GHz) |
Highest Mem Speed 538 DDR |
2-2-2-6 3.5V 1T |
612.3 | INT 3410 FLT 3592 |
INT 7260 FLT 7183 |
77 | 127.2 |
10x272 (2.72GHz) |
Highest CPU/Mem Performance | 2-2-2-6 3.5V 1T |
656.5 | INT 3493 FLT 3697 |
INT 7701 FLT 7605 |
71 | 139.0 |
We reached an extremely stable DDR400 2-2-2-6 at 3.0V with Mushkin Redline XP4000. We then maintained 2-2-2-6 timings all the way to DDR546, which required 3.5V for complete stability. The voltage requirements from 3.0 to 3.6 volts were very linear to Memory Speed. DDR544 is the highest speed that we have ever reached with 2-2-2 timings, but it is only slightly more than the DDR538 2-2-2 achieved in our tests of the OCZ VX. Since both memories use the same memory chips, it appears that Mushkin and OCZ are both doing a similarly excellent job in binning the Winbond chips for their DDR500 2-2-2 products.
The important results here are rows 1 to 6, where CPU speed is kept at 2.4GHZ and only the Memory Speed is varied. The performance differences that you see in that range are a result of Memory Speed only. Since the Mushkin Redline also maintained constant 2-2-2-6 timings across the performance tests, the performance improvements are completely a result of memory speed only. It is not a huge difference in real-world benchmarks, compared to the huge increase in synthetic benchmarks like Sandra, but the increase is real nonetheless.
When you compare results of Mushkin Redline or OCZ VX, you will also see that all 2-2-2 is not created equal. Winbond and the revived BH5 are just a bit faster than other 2-2-2 memory at the same memory speed. Again, the difference is not huge, but where a computer enthusiast is aiming for top competitive scores, every advantage carries weight.
41 Comments
View All Comments
devonz - Monday, May 16, 2005 - link
Ok, maybe I'm missing something, but there seems to be no mention of active cooling being necessary for the OCZ VX modules. Is that correct and why would there be such a difference between the two if they are basically the same chips at the same voltage?Joepublic2 - Monday, May 16, 2005 - link
#18, thats why you use memtest86+, prime95 and some type of looping 3d-demo (or the games you play) to confirm stability. My processor has 25% increase over its nominal clock frequency at the default voltage (overvolting is what wears out dialectrics and causes electromitigration, increasing the frequency alone doesn't put any additional stress on the silicon). My motherboard has a 30% increase over the nominal frequence at the default voltage. And my graphics card has a 16% increase on the core and a %10 increase on the memory, but I don't push it because it already runs so damn hot. And yes, I can notice a difference, although admitably only when archiving, extracting and encrypting very large files and when playing demanding 3d games (the only taxing things that I do with my computer). If I did any scientific calculations, CAD/CAM or rendering, it would make a perceptible difference as well. I haven't ever had a program crash, return a bad result, or had the computer crash, either.Tujan - Monday, May 16, 2005 - link
Buffered,Non-Buffered Ram ?Is this two physically different ram modules.? Certainly would want the Buffered results seen if where to have a choice bearing on performance.
Say you purchase 'Redline ''Buffered''""...or Redline ''unbuffered'"" .
What they do here,physically remove the 'unbuffered modules,and replace with 'buffered modules.
The NF4 supports 'buffered,or unbuffered memories ? This is a specific setting to make sure within the BIOS ?
fitten - Monday, May 16, 2005 - link
#17, there is no such thing as "super stable". It's either stable or it isn't. Having "one memory related crash per week" is not a stable system. Besides, with overclocking, the best result you can get is a hard crash because then you know for sure you've pushed something too far.(Before anyone starts yapping about Windows or the like crashes happening more than stability crashes, my Windows XP boxes - I have 3 that run 24/7 - have uptimes as long as the times between patches that require reboots - granted, that isn't as long as I'd like - or power outages - which we've had two in the past month that were down for longer than my UPSs could keep the machines running.)
I used to overclock everything all the time, then I learned a few things about circuits, CPUs, and digital hardware in general and I also grew up. Stability for me is more important than any 10% overclock that I could ever get. Even if I could get a 100% overclock at the cost of stability, I wouldn't take it.
JonB - Monday, May 16, 2005 - link
On a moderately OC'd DDR motherboard at normal voltages, wouldn't these be super super stable? I know the price is extreme, but some applications (like video or sound editing) need stability first, then speed. If it could stop just one memory related crash per week, the extra money would be worth it.ksherman - Monday, May 16, 2005 - link
is it possible to compare DDR1 to DDR2 in these tests? Ive always heard that they were slower because of the latency, but sometimes here on Anandtech, slower sometimes means 3-5% difference, which isnt much...erinlegault - Monday, May 16, 2005 - link
I think the true test of extreme ram will be once we see some OC tests on the Athlon 64 X2. I'm sure two processors can make use of the higher memory bandwidth.Also, looking ahead. Does anybody feel that the active ram cooling of the Abit AN8 Fatal1ty SLI mobo is enough to cool this ram?
JustAnAverageGuy - Monday, May 16, 2005 - link
#12Welcome :)
Brian23 - Monday, May 16, 2005 - link
I agree with ZeboWesley Fink - Monday, May 16, 2005 - link
#1 & #3 - Our Editing engine is inserting spaces randomly. The original does not have these hiccups. Thanks for listing the locations of the extra spaces - they have been corrected. We are trying to find and correct the problem with the inserted spaces