MA790FXT and G Skill Ripjaws 1600 77724 1.65v Help

Download this:
http://www.tweakers.fr/download/MemSet41b4.zip
Somehow (probably CPU-Z) you're not seeing the whole SPD, including the EPP/XMP profiles, which is where your 'higher speed' settings are stored...

You're right about the spd as I understand it. Bilbat is an expert tho. XMP is not involved since you're using AMD.

But is your RAM running at 1600? You mention the timing, but not the speed.

And try tweaking the voltage up to 1.7 also - 1st thing to try when you can't get advertised latency from the RAM.

Have you previously had problems getting full speed/timing on this motherboard?

Are you runing 2 - 2GB modules? Or 4? Is there other RAM in the system?

Your link didn't work - is this the one?
http://www.newegg.com/Product/Prod [...] 6820231279
Are you sure you got this RAM and not a different RipJaws model? Chk the label
F3-12800CL7D-4GBRH

Are you "Frank" from the NewEgg comments or is someone else with the same board having identical problems?

(I keep editing this!)

The MB specs mention 1666 speed only with AM3 processors, which yours is. But are you running the F5 or greater BIOS rev? F5 is needed for the 955 cpu.

That's true, I've not had trouble with the "other numbers" myself and am frustrated that CPUz doesn't list them all. So is the list in memset the "complete" list of everything that's included in the spd? What about the nanosecond stuff that's often available, I never see that in the spd.

So the thinking is that the other numbers may need to be tweaked to get the main CL=7 number down, right? And need Memset to read those numbers.

And there's no reason to hold off on connecting a computer to the net or installing windows because the timing isn't perfect? Or even the speed? As long as the correct amt of RAM is showing at any speed, you should go ahead and install Windows, update your drivers and get everything running - then work on the other stuff like timing and speed of RAM.

Fair enough. I also participate in the G.Skill forum. You might post there and see what they have in the way of suggestions. I'll chk there and help follow-up if you let me know here you've posted.
http://www.gskill.us/forum/index.php

I think it's odd that the MIT reads the spd as 9-9-9-24 -

did you verify that the modules you have are all labeled as F3-12800CL7D-4GBRH
Do they show the timing numbers and speed on the label?

Double checked my ram and it's sticker says 7-7-7-24, 1600, 1.65v. I tried using Memtest and Memtest 2, both give me an error saying "NOT FOR THIS CPU." Guess it doesn't like my 955.

Could someone give me the XMP info?

Try this:
http://www.memtest.org/download/4. [...] 00.iso.zip
Unfortuntely, no memory manufacterer that I am aware of documents their XMP setings anywhere! This is a major PITA, and I frankly do not know how they've gotten away with it thus far - and with the advent of the i7 platform, these settings become increasingly important; the only way I know of is the tool I posted; there is another, more complete tool here:
http://www.techpowerup.com/spdtool/SPDTool_063.zip
but I am hesitant to have people fiddle with it if they're not extensively experienced, as it's used to write SPDs if you have 'qualified' a particular bunch of DIMMs for a particular MOBO, and want to alter the SPD to automatically 'get' the timings you feel are best for your application - but - you don't want to 'accidentally' modify your SPD - could be fatal (to the further use of the memory - not you )!!

I tried joining tbe g skill forum to talk about this problem there but, they still haven't activated my account.

I'll try that memtest and see what I can get from it.

03, if you decide to update your BIOS, be sure to use the ability to do so within the BIOS - not a Windows program. If you don't have a floppy, you can use a thumb drive to hold the new file.

03, sounds good to me. If you have F6, I can't see it being an issue.

I was just surprised when I found another 790 board recently, shipped with a 955 and running F3.

I've posted a request in the G.Skill forums for ALL spd data on their modules. I'll also post requesting specific full data for your module now.

There is a post there involving a 790FXT that bears reading, even tho its user had slower 1600 @9 RAM.
http://www.gskill.us/forum/showthread.php?t=865

I've adding these posts at G.Skill forum
Specific to this problem
http://www.gskill.us/forum/showthread.php?t=2079

Regarding SPD info in general
http://www.gskill.us/forum/showthread.php?t=2077
http://www.gskill.us/forum/showthread.php?t=2078

I think I may know the reasoning behind this, but I don't like it, or agree with it... I'm pretty sure they simply don't want another seven questions (times however many thousands of users) added to their support load. You've been answering enough questions here to know that, for probably three quarters of the issues, the plain and simple answer is: RTFM! They just don't want users 'dabbling' in these additional settings, assuming that most really don't know what they are for, and aren't willing to learn. A lot of what I try to do here is to (hopefully) 'illuminate' many of the more technical issues in a modern system in terms that can be understood... My belief concerning documentation is that there can never be too much! (I am kind of 'taking care' of fairly elderly parents here - and I do most of the baking - my mom and I laugh about "no such thing as 'too sweet' or 'too much chocolate' " ) This is the major reason that I stick with Intel - I've posted this before:

I know a lot of arcane BS about Intel processors - cause they're the only ones who document everything! If you wanna know how many Lahore pigeons crap on the roof of the Santa Clara fab each year, not only can you find it in a PDF somewhere (but where - that's the skill!) on their web site, but there's probably a three year plan documented to change their feeding habits, so they crap a lighter color, causing the roof to reflect more sunlight, and cut down on the air-conditioning costs... Every time I try to find out something about an AMD BIOS for someone, I see this business about "update AGESA three point five point three point nine point more digits than pi", and I've been randomly trying for months just to find out what 'AGESA' is - bah - no luck! (I hate acronymns anyway - the only one that ever sticks in my head is back from the days when they finally got completely out of hand with 'PCMCIA' - people can't memorize computer industry acronymns!) And you don't wanna even get me started about nVidia! As far as I can figure, nVidia is actually a front company for the CIA/NSA - if you go looking there for documentation, they'll have you investigated to find out why the hell you're looking for their documents!

I think that if they posted this information, along with a cogent, understandable explanation of what they exist for - it would, ultimately, subtract from their support load - but I can't prove it, and I don't know how to get hold of anyone at the manufacturing/marketing level who could effect such a policy... And, as I've mentioned, this info is increasingly important to the new platforms that incorporate the memory controller into the processor, giving infinitely more 'flexibility' to memory configuration; with memory controller hubs, the major function of fast RAM was simply to allow you to raise your system clock past 400 MHz without 'overrunning' your 800 RAM at the lowest 2.0 (1:1 bus ratio) multiplier; now, you can 'diddle' throughput directly - which means you need to know all of the RAM timings, and, at various speeds...

Thanks ever so much for the pointer to G.Skill info - invaluable! It's so good, I had to create an icon for it (their site favicon.ico looks suspiciously like the one from TweakTown...):
http://www.mediafire.com/?g4xon0qtw2y
The best two memory spots I know of are TweakTown's:
http://forums.tweaktown.com/f69/me [...] ios-27283/
and this:
http://www.tweakers.fr/timings.html#tRFC

I'll see about that BIOS adjustment, but I've forgotten what board you have, and figured it'd be easier to just ask than to search...

And glad you liked the G.Skill guides, did you see the DDR2 ones too?
http://www.gskill.us/forum/forumdisplay.php?f=1

So that looks to me like I got slower ram then what I paid for.

Here here! Well said - and another manufacturer whose basic attitude is 'just because you spent your money and bought our product doesn't mean we owe you anything, like good support." Sounds like nVidia - better watch what you say, because the NSA is probably monitoring your comms

Post me a pointer to that discussion - I'll chime in, and add an argument...

I've actually seen this happen - and, unfortunately, I can not remember which manufacturer it was... That's really how I discovered SPD_Tool, for 'rolling your own' - when either the P35/45, or P43 northbridges were released, somebody's SPD data did not work - and their 'service policy' was to have you download the tool, and a new SPD, and flash the thing yourself... And hey - that's not all that much of a bad thing - the world changes, you adjust to it, and move on, day by day. But to deny the information - that is a bad thing! They deal with so many stupid support problems from so many dummies, that they forget that there are some really astute people out there (just look at the linux world, for example) who can lead the way, and help the rest - if only they can get the relevant information themselves!

No - I'm an idiot - and a fairly automated one! To quote Giulio Giorelli (Philosophy of Science at the University of Milan), commenting on a Daniel Dennet book on cognitive philosophy: "Si, abbiamo un anima. Ma e fatta di tanti piccoli robot." Yes, we have a soul. But it's made of lots of tiny robots. I just automatically filter all that stuff out, as my brain is pretty much in a constant state of overload, either from trying to figure out the 'root cause' of the problems, or, from marveling at the cognitive deficits involved! I've actually gotten to the point of avoiding the occasional post, as I've gotten 'wound up' in cognitive deficits before - and it hasn't done me any good...

As for your timing questions:

First, I've never seen that before, or anything even like that! I already had the manual (and I download them [with the exception of the X58/P55 stuff, which I got 'in a chunk', as I knew there'd be problems there!] on a 'case by case' basis), which means I had already answered a question about that MOBO, but I know I never saw that timing section - I'd remember it, as it's so unusual...

Second, it poses a new problem - and, frankly, it's something I should know (or at least, find out)... You know that the DDR (double data rate) tranfers data on both the rising and falling edges of the clock; I often work from 'cycle count' timings, to nanoSeconds, and back - but I've never had to work from the actual nanoSecond timings to counts, directly - the only thing that matters, when converting from, say, 2000 to 1333, is the relative lengths of the cycles. (so the calculations are alway in the form of: { counts * [ 1 / 2000 ] / [ 1 / 1333 ] } ) Thus, what I don't know, is whether, when the BIOS reads CAS4, does that mean four cycles, or four half-cycles (i.e., four data transfers)? But it's a damned good question, and one I'll certainly enjoy digging into!

One thing I'd ask: I know you've seen several of my 'OC' writeups - are you using the <F11><F12> save to/load from CMOS function when you're experimenting? If not - I'd highly recommend it - it saves a ton of time when the board 'burps up' one of your settings, and resets everything to 'stock' (or, even worse, garbles the settings completely, which my BIOS does regularly...)

Memset's 'base' is here:
http://www.tweakers.fr/memset.html
and the place to post problems/bug reports is here:
http://www.xtremesystems.org/forum [...] hp?t=92190
I noticed a beta here:
http://www.tweakers.fr/beta.html
but I think it's the same pointer I've been regularly posting...
Might wanna recommend trying the CPU_Tweaker here:
http://www.tweakers.fr/download/CPU-Tweaker13b3.zip
which seems to be aimed especially at CPUs w/on-die memory controllers - might work for those 955s...

Yeah - it's kind of like 'restore points': "well, if I'd known it was gonna do this, I'd have set a restore point!"

That's exactly the point I try to make here:

I learned (fortunately, early on, during the development of 'structured programming' - mostly from Niklaus Wirth) that if I expend time in carefully choosing my variable names, my code becomes, for the main part, self-documenting. I've spent half of my life doing, mostly, industrial systems (running machines or process factories), where a huge nomber of 'real world' sensors are hooked up, and manipulated by a logic controller; and then the logic controller 'talks' to a windows program, running on a touch-screen, that acts as a sort of 'video game' interface for the operator. These points, or, sometimes, registers, exist in the CAD drawings for the electrical system, in the logic controller program, in the interface program, and, sometimes, in 'glue' logic I might write in VB or C#, to 'diddle' controller data... The first thing I learned was that it was imperative that the names of all these data points be consistent, across the whole shebang, to prevent confusion; the second was that, careful naming was the key to either: not having to spend hours and hours documenting my programming, or, to spending three days in a hotel room five years down the road when I have to service the damned thing, scratching my head and saying "WITF was I doing here?!?" The consistency is one of the reasons I have the four monitor system you've seen - means I can see (and cut and paste, to and from) all of 'em at once. If I'm doing a twenty-four speed Lucas spindle transmission for a boring bar, it's a lot more useful for me to name a limit switch input point "transmission backshaft front gear cluster is forward", than "shift fork A advanced" - in that case, I might need to reference the mechanical drawings and the hydraulic schematics, just to see what is intended... And now, with the advent of object programming, careful naming conventions are just as important as the naming itself - I wanna see at a glance that "that's an object, that's a method, that's a property..." It's a good habit to get into, right down to making yourself up a set of file and directory naming conventions...

Ummm - I keep an index to posts, but mostly not mine - I'll go through it, see if they're vaguely identifiable (there's another place where I need a 'convention ) and send 'em to you. Mostly, when I write up something that will be referenced over and over (as people have, probably three quarters of the time, the same problems over and over...) I save it as a text file in a 'forum posts' directory so I can reference 'em again, at will... After, probably, my third overclock post, I started documenting them at the beginning like this:

and saving them with file names indicating the MOBO and CPU, as I've wound up reposting some of them a large number of times...

Back to the subject at hand - and of interest to the OP.
G.Skill's latest answer:

So I think the answer is that even tho these modules are marketed at 7-7-7-24 they are actually 8-7-7-24 modules. And that's the best that many users can expect from them. The full timing numbers are above and should be used.

I'm not sure that some Gigabyte boards can't use CL=7 in their manual settings. But, in my own BIOS, there are limits to the allowed timing settings, so it's not impossible. We need the user here and the other thread to answer whether they're "allowed" to set the CAS Latency to 7 or not. I would assume they can or they wouldn't say it doesn't work/boot properly at that setting.

OK, final answer from G.Skill is:

So G.Skill feels that the Gigabyte BIOS doesn't allow RAM to run at CL=7. I can't swallow this and would love to get info from a user who IS running 1600 @ 7. And clearly slower RAM can run at 7 anyway.

And I still haven't gotten any confirmation that the SPD for this RAM shows 7's for the 1600/800 column.

I've been reading this thread, and a related one (http://www.tomshardware.com/forum/264412-30-really-exact-timings-screen-freezes-help), with much interest, as I'm having similar struggles with my AMD/Gigabyte/G.Skill homebuild. Thanks much to all of you for the valuable pointers!

Like the OP (and the OP of the 2nd thread), I cannot get memset to run on my computer: it gives the "Unsupported CPU" message. CPU-Z runs fine, but only gives 8 out of 16 figures that I need to fully configure the RAM timings and voltage.

[Edit:]
So here's another tool I found--you old timers have probably heard about it--which does show the full SPD values for Phenom II with DDR3: "CPU Tweaker 1.2", which can be downloaded here: http://www.tweakers.fr/cputweaker.html .

I only noticed two weird things in the CPU Tweaker output:
(a) it shows my current CAS# as 7, when it shows in BIOS as 9.
(b) It only lists one of the four "tRFC" values. There's one entry for each memory slot; I have two cards, and BIOS shows the same value for each (110 ns). CPU Tweaker only lists the value for slot #1. Also there are no tRFC values in the SPD section...oh well, nothing's perfect!

Besides these odd glitches, everything reads as expected.

I don't have image hosting set up, so here's the SPD values in plain text from CPU tweaker for DDR3-1333 with 7-7-7-18 timings, which is the "XMP1" column:

Clock = 666 MHz
CL (CAS#) = 7
tRCD = 7
tRP = 7
tRAS = 18
CR = 2T
tRC = 30 (Row Cycle Time in Gigabyte BIOS)
tRFC = 60 ( Doesn't show in BIOS...seems like it ought to!)
tRRD = 4 (RAS to RAS Delay in BIOS)
tWR = 10 (Write Recovery Time)
tWTR = 5
tRTP = 5 (Precharge Time)
tREF = 7.9 micsec (??)
voltage = 1.60

So hopefully this info will be useful to others with this same RAM and motherboard! I'll post again once I've tried out the new timings.

My system:
Phenom II X3-710, Gigabyte GA-MA770T-UD3P, G.Skill DDR3-1333 2x2GB (F3-10666CL7D-4GBPI),
2x640 GB WD "Green" edition (7200 rpm) in "fake-RAID" via Gigabyte driver [I know, fake RAID is cheesy, but it seems to be working suprisingly well!], Asus/nVidia GTS 250 512 MB, Antec Earthwatts EA650 PSU.