Jump to content
Corsair Community

Where are stats on new XMS 4400?


dogin

Recommended Posts

OK, I really love the 3500LL pro ram, but going to put it in my FX53 system and try for more performance on my FX57 system with the new XMS 4400 pro ram that will be here the end of the week. Really would like to know a good starting point to get the full benefit (ie running 558 DDR @ 2.5-3-3-7 a good starting point? At what v's 2.7??? and it looks like the HT would have to be x3 or is it better to run an HTx4 and lower the ram speed? ). Can't find the stats under the sticky post yet. Is there anything posted that explains the speed settings this ram can do and the timings??

 

Well, found part of answer on the Corsair web page PDF 3-4-4-8 at 2.8v's and other review sites. The command rate is not listed and there are various reviews that say it is Tx2???? Two of the reviews go on to say they were able to run it at DDR500 3-4-4-8 with Tx1. I hope so or the results of the Clocked 3500ll will beat it??? It also looks like Corsair only tested it on one mobo (not identified), but both testors used a Lan party mobo?? Hope can get the same results on the Asus (looks like it didn't work on some others??)

Link to comment
Share on other sites

PC4400 memory is rated to run up to 275MHz and some boards/systems/power supplies, etc just aren't up to it. Therefore, it's always going to be hit or miss with such a large overclock.

 

However, the command rate issue is well known.

 

The command rate is much more a controller setting than it is a memory setting and the ability of a memory controller to run a 1T command rate is inversely proportional to the frequency at which it's set. Thus, the faster you make it go, the less likely you'll be able to run a 1T command rate.

 

Some systems' cut-off is around 250MHz (PC4000) and some is higher, but some are also lower... but a little extra voltage can always help if you watch your temperatures!

 

For the record: XMS Qualification and Testing

 

SDRAM CAS Latency: 3T

SDRAM RAS to CAS Delay (tRCD): 4T

SDRAM Row Precharge (tRP): 4T

SDRAM Active to Precharge Delay (tRAS): 8T

SDRAM Bank Interleave: 4 Bank

SDRAM VDIMM: 2.75

All other settings are motherboard defaults.

Maximum recommended VDIMM: 2.9

 

PLUR

CK

Link to comment
Share on other sites

PC4400 memory is rated to run up to 275MHz and some boards/systems/power supplies, etc just aren't up to it. Therefore, it's always going to be hit or miss with such a large overclock.

 

Thanks CandyKid. Knew the memory controller was a limiting factor on my 3200ll, but somehow thought the 4400 may take it a little higher with Tx1 being possibly easier on the controller. That doesn't appear will be the case. Guess I will try a higher clock at Tx2 to see if I can get any better performance, but it looks like the 3200ll may be the best option???? The 4400 is due here on the 24th so I can try to see if it was a waste of $'s.

Link to comment
Share on other sites

You'll really just have to test you rsystem's bandwidth and real-performance with programs like Everest, Sandra, and 3DMark to tell which is 'better'.

 

275MHz at 2T may be better than 216MHz at 1T, I'm not sure, but let me know!

 

PLUR

CK

Link to comment
Share on other sites

That 3500ll may just be the top ram as it runs Tx1 3-4-4-8 2.75V's @ HHT of 250 (DDR500) and a Sandra 7107Int buffered result. Thats with the FX 57 at 3.0GHz. To get 3.094GHz have to drop the 3500ll HHT. Maybe that 4400 with a little V push may go over the 500 at Tx1 so can run the CPU higher too??? Saw where one review was running the 2048 4400 on my board at 546 Tx1 but they did not give the v's they were using nor did they mention if they had a CPU with the better performing memory controller. Like you said it is hit or miss and I am hoping to hit a little higher on this one (will try the Tx2 option too).
Link to comment
Share on other sites

Well, The 4400 running at the same speeds as the 3500ll OC'd post just slightly higher benches in Sandra. The 4400 runs at 2.5-3-3-7 Tx1 2.75V's up to at least 238 HHT with FXf7 at 3.094GHz (which is the highest I could run the CPU and stay stable). Speeding the ram up to 250 and above required Tx2 settings and I never reached a stable Tx2 over 250 HHT that would beat the Tx1 performance under 250. Since I gained slightly higher bandwidth with the 4400 than the 3500ll OC'd at the same speeds and same settings, am happy with the 4400 (even though could not reach what I had hoped for). Note: I adjusted the HT multi to keep the HT at or under 1000. Increasing ram v's did not improve anything. This mobo and CPU appear to be the limiting factor as I could not run 250 and over at Tx1 which some reviewers could? Did notice that increasing the CPU from 1.45V's to 1.4625 to get from HHT 233 to 234 the bandwidth dropped slightly before heading back up as the HHT was increased. To me that looked like a point of diminishing returns for the memory controller???????????
Link to comment
Share on other sites

  • 2 weeks later...
Well just read a review where an Lan Party mobo and a CPU with an exceptional memory controller pulled off an HHT of 280 on these sticks. Maybe time to check out another mobo to see if there is any improvement????
Link to comment
Share on other sites

I'd chaulk that one up 99% to the CPU, not the mainboard. It must've had an amazingly well made core / memory controller / etc to pull that off.

 

PLUR

CK

Guess will save the mobo experiment $'s. All the time I have been building, have never lucked out with the top performing chip. Gets discouraging.

Here is a link to the review:

http://www.pureoverclock.com/article35.html

Link to comment
Share on other sites

I'd chaulk that one up 99% to the CPU, not the mainboard. It must've had an amazingly well made core / memory controller / etc to pull that off.

PLUR

CK

 

I have to somewhat disagree with CandyKid, here - although at present it is theoretical as I haven't had time to really test things out myself yet.

 

The DFI motherboards have memory adjustments NOT AVAILABLE on Asus boards (fine though they are).

 

This is even more true on the Expert than the older series (Ultra-D through SLI).

 

And the Venus has very much upgraded caps allowing even faster signals through the board. ( I can atest to vastly better RAM compat. even at reduced voltages! :biggrin: )

 

I understand the newest ATI chipset DFI has super tweeking also, but not the polymer caps.

 

If you have a friend with a DFI board you might see if you can swap for a while. Youv'e already spent a TON on RAM - might as well get the most out of it.

 

Be advised - these DFI boards take a LOT of trial-and-error then start again and you can't just rely on what worked for the other guy.

 

Good Luck! :sunglasse

Link to comment
Share on other sites

"Be advised - these DFI boards take a LOT of trial-and-error then start again and you can't just rely on what worked for the other guy."

 

Absolutely true. This is the cause for two things:

 

1) Great OC's.

 

2) Horrible compatibility.

 

Number 2 here isn't really a hardware conflict more than it is a misunderstanding of all the settings... DFI boards when properly set can do amazing things, but they do tend to give the user enough rope to hang themselves, if you know what I mean.

 

Give it a good try though, they can be great. I think the guys at DFI Street have a few guides (or should have!) to tell you about most of the settings.

 

PLUR

CK

Link to comment
Share on other sites

  • 3 weeks later...

Appreciate the input and advice. This system is running so good now (ie Oblivion, FEAR etc running at max settings), that going to just play games for awhile. Later will go back to the OC trial and error and try the DFI board. This ocing addiction just removes you from using the PC for anything else. Looked back at the hours/days spent in tweaking, stability testing etc and think I need some gaming time for awhile.

Thanks again!!

Link to comment
Share on other sites

Archived

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

×
×
  • Create New...