I'm under NDA on any specific changes outside of the 4 holes, but I can say there will be some changes:)
Randi:D
Printable View
I'm under NDA on any specific changes outside of the 4 holes, but I can say there will be some changes:)
Randi:D
when is this NDA gonna fall? :D
i hope these changes will be in the xtreme enthusiast-mobo way!
Randi, thanks. I see 3 tall caps just on the south edge of the ziff and a couple between the ziff and power connector. They are all dark blue/black with a stripe of gray. Can I assume these are the ones we are talking about? Could this explain why my mobo struggles above 190 fsb?Quote:
Originally posted by MrIcee
Nichicon caps are silver...Rubycon dark blue/black:)
Randi:D
Also, there are 3 more further south of the ziff that are silver.
Quote:
Originally posted by bigdawginva
Randi, thanks. I see 3 tall caps just on the south edge of the ziff and a couple between the ziff and power connector. They are all dark blue/black with a stripe of gray. Can I assume these are the ones we are talking about? Could this explain why my mobo struggles above 190 fsb?
Also, there are 3 more further south of the ziff that are silver.
check out the 2 big ones between mosfets and agp ports!(in between there's the HIP voltage regulator )
mine's are the silver ones and the mobo oc like a champ
Ok, thanks. If you're referring to those then all 3 of those are silver so I suppose that's not the porblem.
I'm giving up for the night. I've tried a 256M stick of TwinMOS/Winbond BH-5 and a 256M stick of Samsung CTL-TCB3.
No matter what I do, my new KD7-E won't run DDR400, only DDR333.
I've done the following, and always started out resetting CMOS for 5 minutes followed by setting "Optimized settings" per Mr Icee's recommendation, and the first post after setting optimized shows DDR400, but I've done the following without success:
Set 200 FSB and change NOTHING on the first post from Optimized, TwinMOS and Samsung.
Set SPD on the TwinMOS/Samsung, and 200 FSB.
Set SPD on the TwinMOS/Samsung and walked up in 10's from 133 to 208.
Set 200 on the TwinMOS/Samsung, and 200 FSB.
Set 133 on the TwinMOS/Samsung, and walked up in 10's from 133 to 208.
Set 166 on the TwinMOS/Samsung, and walked up in 10's from 166 to 208.
Repeated the above after a new flash of the C8 BIOS. Twice.
No matter, it always posts showing DDR333 at 200 FSB, and gives PC2700 bandwidth in Sandra (2700 MB/S Float). :(
The Samsung is the same stick that gave me 3200 at 208 with my original KD7-E.
Can you guys help out here? If this board won't do DDR400 I'm afraid it's the last straw for me and ABIT, this board acts like it's been purposefully crippled from the factory. Yah, I know, I'm paranoid - but I was so very impressed by the potential of my first board that this is REALLY getting to me.
sjohnson:
Your first board must have been a glitch or something. I've had a number of these KD7-Es pass through here now, and none of them would run DDR 400 at any speeds above 165mhz fsb. Starting with a FSB of 166mhz, the memory and FSB are automotically run synchronously.
This is really no different than the original KD7 (KT400) and is limitation of the chipset.
Are you sure that your memory was running asynchronously at speeds above 166mhz on your first board? Could it be differences in your basic memory timing settings that account for the difference in Sandra scores.
For comparison purposes my Sandra float is about 2760 at 195mhz, 2832 at 200, and 2903 at 205. This is at CAS 2, 2,6,2 2T.
You're missing my point, musta been my lack of sleep and frustration when I tried to explain. My board won't run synchronous above 166 FSB. At 200 FSB, I'm still running asynchronously. Sucks to have the ability to run well in excess of 200 FSB but your memory is stuck at 166.
(Edit) - rereading your post, I'm confused about what you're trying to say - if the KD7-E won't run DDR400 at any speed above 165, how can it run synchronously at 200?
I never run asynchronous by choice.
This is from my original 7-e. I think it shows that I was running synchronously on the first board. My examples above were to show how I tried to force synchronous behavior above 166 FSB, without success, on my new board.
If you run the board at speeds below 166mhz FSB, you can select the memory to run asnychronoulsy at DDR 400.
A good example is for the avererage joe who just pops in his 133mhz processor (133 double pumped), and sets the memory to run at either DDR333, or DDR 400. In this instance, the mobo will run the FSB at 133mhz and the memory asynchronoulsy at 166 or 200mhz.
Performance gains are just about nil when doing this, though, as memory latency penalties for running in asynchronous mode kill the gains...not to mention the CPU bus bottleneck at 266mhz (133 x 2).
You say that at 200mhz fsb your memory is still running aynchronously. What speed is the memory running at when your FSB is at 200mhz? What does Sandra say about the memory bus speed stated in mhz?
sjohnson:)
When you set FSB AND memory to 166Mhz...you are in Sync. It will post from then on as DDR333....no matter where you set the FSB. Even if you set the FSB to 200...the memory will post DDR333. The only time I've ever seen it post DDR400 is when you run ASYNCH and FSB is 133 or 166...and the memory is at 200Mhz.
You cannot pay attention to the posted DDR333 speed(or the 166 speed in the memory section of the bios). Go into Sandra after you've done as I've explained above. For giggles, set your multi to 10....FSB to 166 and in the memory section the memory to 166Mhz. Mind you from here on it will show the memory at 166 in the memory section and DDR333 at post. Reboot and go into the bios...set the FSB to 200Mhz and reboot. Boot into Windows and run Sisoft Sandra's Mainboard configuration. You will first see FSB listed, then scroll down, then memory and it's speed. They should both be showing 200Mhz.
Give it a shot and let us know how you make out:)
Randi:D
PS. The reason it shows DDR400 after setting optimized defaults is that optimized it's running async 133/200 or 166FSB/200Mem. You WILL NOT see DDR400 at post EVER running in sync mode...it always shows DDR333 even when you are running 200/200.
BOTTOMLINE: Set Memory to 166Mhz and leave it. Set FSB to 166Mhz.From there raising your FSB ALWAYS raises FSB and MEM
syncronously all the way to your hardwares limit. Use Sandra's mainboard configuration module to confirm
Looking at that Pic, you need to scroll the Sandra box down to see the memory speed in mhz. What you can see there is only the FSB speed.
I'm quite sure that the pic above was running synchronously based on the scores you have.
On the new system, run the same bench and scroll down and check the memory speed.
BINGO! Thanks for helping explain that Randi! I wasn't sure why he was thinking he was running at DDR333 when he was at 200mhz FSB. I didn't even think about the post screen....I'm so used to ignoring it now:)Quote:
Originally posted by MrIcee
sjohnson:)
When you set FSB AND memory to 166Mhz...you are in Sync. It will post from then on as DDR333....no matter where you set the FSB. Even if you set the FSB to 200...the memory will post DDR333. The only time I've ever seen it post DDR400 is when you run ASYNCH and FSB is 133 or 166...and the memory is at 200Mhz.
BOTTOMLINE: Set Memory to 166Mhz and leave it. Set FSB to 166Mhz.From there raising your FSB ALWAYS raises FSB and MEM
syncronously all the way to your hardwares limit. Use Sandra's mainboard configuration module to confirm
Yeah...the initial post screen at DDR333 confuses alot of people. It's because you locked the memory in sync at DDR333 by setting it at 166Mhz. It's a shame it doesn't change and adjust to reflect it's ACTUAL speed as you scale the FSB upwards. This anomaly has been the frustration of many setting up KD's since their introduction. Using the mainboard configuration module in Sandra is the easist way to confirm you in fact are sync and running the speeds you've set in the bios:)
Randi:D
Same CPU, same memory (Samsung CTL-TCB3), same memory timings, same voltages, same HD and peripherals. The only change is the the board and it's the same make and model, KD7-E.
From my list, sorry it was so long:
"Set 166 on the TwinMOS/Samsung, and walked up in 10's from 166 to 208." So, I did set 166 and then ran it up. Still showing scores indicative of running at PC2700 instead of PC3200. What gives?
(Appreciate the really quick responses guys. Sorry, but I have to head out to work. Will pick this up tonight. Again, thanks - this is such a nice board that I'm really frustrated that I can't seem to run anywhere near this Sammy or this TwinMOS bandwidth on my new KD7-E).
Quote:
Originally posted by sjohnson
Same CPU, same memory (Samsung CTL-TCB3), same memory timings, same voltages, same HD and peripherals. The only change is the the board and it's the same make and model, KD7-E.
http://www.xtremesystems.org/forums/...=&postid=85726
What FSB speed is this at?
I'm out the door, will repost tonight. (You'd have to take my word for it at this point), it's identical to the SS showing 3200 on the original, 2.5-2-2 1CMD, Enhance/Enhance/full aggressive/208 FSB.
Your first SS shows the CAS setting at 2 in Sandra, not 2.5. That's going to be your major difference:)Quote:
Originally posted by sjohnson
I'm out the door, will repost tonight. (You'd have to take my word for it at this point), it's identical to the SS showing 3200 on the original, 2.5-2-2 1CMD, Enhance/Enhance/full aggressive/208 FSB.
You also didn't run the mainboard configuration screen as I asked and checked the FSB and memory Mhz settings...that will confirm its running in sync and at whatever you have the FSB/Mem set at. Bandwidth tests are not conclusive at all, especially as Mike pointed out you are running at cas 2.5 as opposed to cas 2 in your first post
Please run the Sandra mainboard configuration module later today and give us the lowdown on the FSB amnd memory speeds from it.:)
Randi:D
I would not trust Sandra, look at these screenies(first one is CAS2,second one CAS2.5):
http://home.tiscali.no/kartoffel/sandra180cas2.jpg
http://home.tiscali.no/kartoffel/sandra180cas25.jpg
CPU-Z also mistakes CAS2.5 for CAS3. Maybe its a mobo problem? Anywas, you can also notice that the performance hit from running CAS2.5 opposed to CAS2 is not big. Are you using the same BIOS? I think its strange that Abit has replaced the caps but not changed revision. Also, from my experience with Nichicon vs Rubycon on the KX7-333, the Nichicon board smoked the Rubycon board in stability, OC ability and voltage stability. Strange problem sjohnson :(
damn thats a mess, wonder why :confused:
I would think if he could post these two shots for us that could confirm either way.
OK gents, just got home from work, long day and no 'net access. I'll fire up the 7-E and see if I can't respond to each request.
Sorry Mr Icee, I was incredibly rushed this AM, that's why I haven't yet responded to your requests for more info. I will, though! :) Spoiler, Sandra will show the memory running at 208X2 at 208 FSB.
So much else is fubar with sandra's reporting that I assumed it was wrong as well. My memory bandwidth on this board is currently well below what I know the memory can run at the same FSB on both of my KR7A and KX7. However, I trust Mr Icee and mdzcpa'sknowledge on this board - if they say it's running 200 I did NOT intend to argue the point and apologize if my posts seemed to do so. Perhaps the bandwidth experienced with my first 7-E was a fluke, one of a board destined to be RMA'd.
My plans, after this ardent help and time taken to explain will be to gently burnin this new board and see what might happen.
Got to help my son with homework now, will post the requested infos later tonight. Again, thanks to all!
sjohnson:
No need to apologize:) Post the info when you can and we'll see if we can come up with some ideas to help.
I would have to agree now with you guys about Sandra being all borked up on this memory reporting. I guess I nevered looked hard enough at it before.
Not only does it refer to regular "SDRAM" and the wrong efficiency % as mentioned somewhere earlier in this thread, all the settings are off. More like mislabled maybe.
I think it's a Sandra issue based on my experiences benching it and the close similarities to my KD7-RAID memory benches. The benches I'm getting are almost exactly what I could get from my KD7. Which makes sense to me as both boards use the same memory controller.
Although...sjohnson's problem does puts a wrinkle into the mix until it can be fixed/explained.