KD7-E initial impressions!

[mdzcpa]

Hey SPQQKY:

Up to 200 fsb for you it seems

[Nohto]

Hey mdzcpa what is the highest FSB that you've achieved 3DMark proofed? So far I am burning in a second 256Meg Corsair 3200 platinum for my Nforce board and I am up to 195 max timings at 1.85 volts.

[DuraN]

I have got this 3DMark high fsb:

http://service.futuremark.com/compare?2k1=5348582

But this is my ram stopping me,can do 215 mhz windows (prolly 210 3DMark) but the bandwidth is bad due to slow timings

i believe SlimHitman has achieved a 212mhz fsb and even ran the full test

[Nohto]

What were his voltages and timings? Also do you know if he was using a 333 or a 266 processor?

[mdzcpa]

Originally posted by Nohto
Hey mdzcpa what is the highest FSB that you've achieved 3DMark proofed? So far I am burning in a second 256Meg Corsair 3200 platinum for my Nforce board and I am up to 195 max timings at 1.85 volts.

As it sits now with the R9700 Pro in there, it will loop 3DMark2001 at 202mhz indefinitely, and make solid bansi runs that can get through the bench once or twice up to 205mhz. Unfortunately the Radeon is FSB sensitive so that's where the fun stops.

Memory timings at that speed are at CAS 2, 4 way, 2,6,2 @1.85v with 2 256 sticks of XMS 3500C2.

If I stay out of 3D apps the system will boot and run Windows just fine up to 218....Prime stable. I am contemplating putting my Ti 4600 back in there just to see how high things will go FSB wise in 3Dmark. But I know my overall 3DMark score would drop so I haven't been eager to do that just yet.

[Nohto]

What about the last set of timings :
Dram Access
Enhance Dram Perf.
Dram Command Rate
Write Recovery Time
Dram tWTR

What processor are you running? 333 or 266?

[mdzcpa]

Dram Access: 3T (default)
Enhance Dram Perf: Enabled
Dram Command Rate: 2T with 1.85v, 1T with 2.95v up to 205
2T above 205 with 1.95v

Write Recovery Time: 3T (default)
Dram tWTR: 3T (default)


Processor is a 333mhz XP 2700.

[Nohto]

Thank you good sir I appreciate it.
Nate

[mdzcpa]

Anytime M8

[bigdawginva]

So far rock solid at 190*11.5 @1.85v but 195*11 won't boot into windows even with the most relaxed memory settings. I'm guessing it's the 9700 but I'll try this old GF3 to make sure.

[sjohnson]

Sorry for the delay in reply, just got back from a trip.

Newegg p*ssed me off, they won't give me a full refund - I have to pay 15% restocking fee on a broken board.

I don't know why, but they don't want to cross-ship on a "repair' RMA - and I don't trust them on a "repair" - I want a new board not a refurb and they don't seem to want to assure me I'll get a new board from them.

So, I'm taking a 15% plus shipping loss and getting my money.

Don't get me wrong, Newegg is great for stuff that works but I've had zero satisfaction with them on RMA stuff. The last RMA I got from them was on a refurb and the received replacement was DOA. Sent it to ABIT, cross-shipped and was ultimately happy.

Another KD7-E sounds appealing but I may just chicken out and get a KX7. Has anyone yet gotten to 220 FSB on the KD7-E, at least stable enough for a WCPUID and Sandra SS?

[SlimHitman]

I'm still stuck at 214fsb if i try 215 with udma on it scrambles my OS.I just tried setting it to pio4 and it just reboots instead of scrambles.

Wish they would hurry with the kt400a.

That sux about Newegg shafting you with the 15%.

[sjohnson]

Ow! Before my kd7-e started going south on me, I had it up to 218. No screenies, I was pushing for higher FSB and after a 12-hour sandra mem burnin at 218 I tried for 219 and my problems began It *was* sweet while it lasted tho

[mdzcpa]

I'm not sure whether my KD7-E could get there or not. My R9700 Pro begins holding me back above 205mhz

Still, though, this is the most silky smooth 200+mhz board I've ever had. It has yet to bsod, freeze, or bomb out in Prime/3DMark...and it runs 24/7.

Man...I hate sounding like a cheerleader about this board. But it's just worked so well...for me anyway

[sjohnson]

Agree with mdzcpa. A nice, smooth board.

I don't know why mine went bad but think that the NB went sour since both memory performance and Max FSB went way down prior to the controller failure. In fact, it seems likely that the NB either influenced or caused the IDE failure.

I was aggressive in working it up to high FSB (less than a week from receipt to 218 FSB), perhaps a long, gentle burnin would have been more productive.

[sjohnson]

OK, I liked the board so much that I bought another. Newegg was OOS plus the RMA left a bad taste in my mouth, so

$84.99 *shipped* at googlegear: http://www.googlegear.com/jsp/Produc...uctCode=240095 - almost recoups my lost newegg $$, at least makes it less painful

[bigdawginva]

Since I'm still struggling to get over 190 fsb even with removing the PCI cards, I thought I'd try a different approach.

I realize Sandra is unreliable but it indicates my AGP and PCI buses are running 96 and 48mhz, respectively.

Clearly, this would indicate it's seeing a 1/4 divider and not 1/5. Can I assume the 1/5 divider is indeed operating?

[sjohnson]

I haven't tried Sandra 2003, but 2002 has problems reporting correctly on the KD7-E.

I got the same relative AGP and PCI speeds on mine that you're getting, plus the added bonus of Sandra "seeing" SDRAM instead of DDR, so my memory bandwidth efficiency was typically 192% - don't trust Sandra's reports on this board until they put out a version that supports it.

[N8]

Originally posted by bigdawginva
Can I assume the 1/5 divider is indeed operating?

Indeed you can. Sandra hasn't reported the KT333 divider properly for some time now

[bigdawginva]

Thanks. I never had a need to check with my KX7-333R so I didn't realize this had been an ongoing Sandra issue. I verified by switching to 1/4 and it wouldn't even boot so that's when I confirmed the 1/5 was indeed working.

I can boot 195*10 and 200*10 all day long but will never get into WinXP where it just hangs. That's even with the PCI cards out. My only other thought is the NB, caps and/or mofsets are too warm to get past 190. I'm not going to add more fans just to cool the mobo as I'm quite pleased with this new, almost silent running rig as it is.

Again, thanks for everyone's input. My overclocking buzzard's luck continues. At least it can run 15k without any video card overclocks.

I'll going to sell this new rig and move back to Intel in the coming weeks.

[SPQQKY]

I have been dying to give this board a run for it's money but have yet had the time. I probably could have yesterday, but the Corona's and Rumple Minze the night before took it's toll. Hopefully this weekend I can give some results. So far it has been rock stable for my everyday use. My 2400+ has been running as a 2800+ for 4 days now (166x13.5). Not sure what this Radeon 9000 will do either, my last one did 331/300 before it died.

[bigdawginva]

Here are my two benches 17014 16618

The second bench (KD7-E), although run with different mobos, has the CPU clocked higher (190 fsb) than the first (KX7-333R) but with the same video card settings (380/340).

I realize everyone's more interested in Oppainter's 21k efforts but I could use some advice.

1. Would the mobos cause the 400+ mark difference?

2. What drove the lobby scores so low?

3. What components and/or settings affect Lobby FPS?

[mdzcpa]

Hmmm...that is strange.

When I was considering doing a KD7-E mobo review, I compared the KX7, KD7, and KD7-E bench scores (using identical configurations). And I pretty much had all the scores clustered together. None of them lead the pack by any significant margin. Based on that, I beleive there is something else definitely going on with your systems...something set up between them differently.

I am no 3DMark expert, but I can throw out a few ideas to consider as to why you have materially different scores. You would need to make sure these items are identical between the test systems.

Memory timings. These can really effect the score at higher fsb speeds too.

AGP settings such as APG aperture and Fastwrites.

Video card overclock, core and memory, if any.

Vid driver tweaks such a Vsync, refresh rate fix, and all the performance vs. quality settings.

A fresh install of the same OS with the appropriate service packs, 4in1 drivers, and vid card drivers.

Video card driver version.

These are the biggies anyway. It's very hard to have an apples to apples comparison unless you run these boards together, one right after the other, with the exact same hardware (except the mobo), BIOS set up and OS/driver install routine. they must all be exactly the same. Easier said then done sometimes.

Hope this helps I'm sure a real 3DMark expert might be able to help you more.

[bigdawginva]

Thanks for the awesome feedback and advice. First, although I can't run it as high as some, I can run it higher than any other mobo I've ever owned. It's a great mobo. I've included my comments below.

Originally posted by mdzcpa
Hmmm...that is strange.

When I was considering doing a KD7-E mobo review, I compared the KX7, KD7, and KD7-E bench scores (using identical configurations). And I pretty much had all the scores clustered together. None of them lead the pack by any significant margin. Based on that, I beleive there is something else definitely going on with your systems...something set up between them differently.

I am no 3DMark expert, but I can throw out a few ideas to consider as to why you have materially different scores. You would need to make sure these items are identical between the test systems.

Memory timings. These can really effect the score at higher fsb speeds too.
I think this is a biggie as I have it set on TURBO so I need to go to MANUAL and set them more aggressively.

AGP settings such as APG aperture and Fastwrites.
They were the same between mobos according to my notes.

Video card overclock, core and memory, if any.
They were the same between mobos according to my notes

Vid driver tweaks such a Vsync, refresh rate fix, and all the performance vs. quality settings.
They were the same between mobos according to my notes.

A fresh install of the same OS with the appropriate service packs, 4in1 drivers, and vid card drivers.
They were the same between mobos according to my notes.

Video card driver version.
The video card drivers (6193 vs 6218) and I believe this may have something to do with it as well.

These are the biggies anyway. It's very hard to have an apples to apples comparison unless you run these boards together, one right after the other, with the exact same hardware (except the mobo), BIOS set up and OS/driver install routine. they must all be exactly the same. Easier said then done sometimes.

Hope this helps I'm sure a real 3DMark expert might be able to help you more.

[SPQQKY]

A little hard to tell without the full tests on 3DMark. The fill rates would help determine if the card is somehow suffering. I had a problem with decreasing scores on a rig that was exactly the same, only difference was the card was getting tired and couldn't perform at the same clocks, that was right before it died. But then Lobby is more cpu dependant, so I can't say that is your problem.