Asus Striker Extreme

[rt182]

Hey everyone
Just got a new Asus Striker extreme cause my GA N680SLI DQ6 just wasnt doing it for me i just have a few questions for all those users who have one or know about them. which bios should i be using?? are the any beta bios, im currently running a QX 6700, XFX 8800 GTX XXX and 4GB of corsair dominato PC 8500 C5D
Basicly anything you can tell me
Drivers, any warning i should take ??? running at 1067 with my RAM right now everything is still stock 800 mhz ect. voltages are still all on normal aswel. im still getting use to these new bios compared to the gigabyte ones they are a bit different
anyway any other info or advice gladfully appereated
Thanx Rob

 

[McCartney]

You will have problem with any Nvidia 680i board and 1066mhz memory.
I've gone through 4 strikers and 2 EVGA 680i's
You should return it and hope 780i is our fix. I am doing the same so my 2 cards are collecting dust ;(

 

[BonzaiDuck]

rt182 --

I'll explain briefly my experience with the STriker Extreme -- hoping that I don't get flamed by friends here who didn't like it so much.

I bought one in March '07 for an E6600. It had the 1004 BIOS version, and there were no problems whatever.

But for the reasons of wanting to improve onboard fan-control, and just out of obsessive-compulsive habit, I was eager to flash the BIOS with every new revision release. This was all fine, until I tried to flash version 1301. I cannot tell whether it was the BIOS revision or me -- but I'd forgotten to reset the CMOS before flashing. After that, it wouldn't boot.

A little cool thinking in aftermath of panic, and I found these folks:

BIOS Man

They'll flash a BIOS chip for you and send it in the mail for $25. The $25 buys you a year's subscription to their service (something like that), so that all you need to do is send them your spare BIOS chip and they will flash it and return it. There were a pile of people who had trouble flashing version 1303 or 1305 (can't remember precisely).

If I were to recommend anything, I'd say DO NOT use the Windows-based ASUS-Update program to flash your BIOS -- I think I had much better luck with the EZ-Flash feature built into the Striker BIOS. You could also use the DOS_based flash utility, but using EZ flash is "EZ-er."

If you're running v 1301 or 1303, make periodic checks at the ASUS forums and check for new BIOS versions on their web-site. For a Q6600, I'd say -- stand pat with either of those versions until you plan on replacing the CPU with a Penryn.

Also -- there's no guarantee that Striker Extreme will ever be compatible with the Yorkfield Penryn -- it's currently only cert-ed for the Wolfsdale dual-core version.

For over-clocking it, I advise first unhinging the "link" between FSB and DDR, and strive for a 1:1 divider. Go for a 1,334 FSB and DDR=667 -- that should get you to 3.33 Ghz -- easily. From there, you can test the limits of the RAM by tightening down the latencies (see if you can get CAS = 3), and then shoot for a 33% OC by loosening them slightly and setting the RAM voltage just below the warranty limit (is it 2.2V?)

You may want to twist up the CPU_VTT voltage and 1.2VHT a notch, possibly two, and likewise with the NB voltage. I'd leave the SB voltage fixed at 1.5V, and set your PCI_E speeds to 100,100, 100 and 200. Turn off all spread-spectrum, Intel SpeedStep -- that is, disable everything except the AV bit-setting in BIOS.

Here -- take a look at this -- although there are also some good reviews from last year and early this year with "how to" details:

'EVA' is a guy in Australia

 

[BonzaiDuck]

Just for the record -- I'm now running a Q6600 @ 3.204 Ghz, Crucial Ballistix DDR2-1000 @ DDR2 712 Mhz, 3,4,4,8.

You really should not have any problem with a QX6700.

 

[rt182]

Originally posted by: BonzaiDuck
Just for the record -- I'm now running a Q6600 @ 3.204 Ghz, Crucial Ballistix DDR2-1000 @ DDR2 712 Mhz, 3,4,4,8.

You really should not have any problem with a QX6700.

wow thanx thats quite a in dephth answer, thanx tones !!!!, just one question those PCI_E speeds, is that lowering them? 100, 100,100 and 200 ?? is there a reason for this and also i went onto the asus site and i think i saw the latest bios were the 1305? yeh so i shouldnt use this? or should i
anyway thanx again i really needed help with all those voltages cause i have no idea.
btw that Q6600 what FSB are u running at.. on my old GA N680 i couldnt really get past 330???? mayb i was doing something wrong anyway
thanx again
=D

 

[The-Noid]

Gigabyte had an older version nb.

You can get upwards of 1000+ mhz by increasing CPUVTT to 1.40, FSB Termination to 1.40, Southbridge (ICH) to 1.60v, and NB to 1.40v. I was running around 1066 C5 x 4 dimms fo r quit ea while on striker. It also helps if you set DDR2 Termination +.15mv

 

[BonzaiDuck]

My pleasure, and Yorxxy also put some good info in your lap.

The PCI_E bus speeds, as set to these absolute values, are equivalent to "fixing" the AGP/PCI ratio on older motherboards. Otherwise, if you overclock the system's FSB (which is exclusively essential to OC'ing other CPUs below yours in the product line), the speeds increase proportionately, and will either cause instability or permanent damage, to data, hardware or both.

In case you didn't already know -- it is EXTREMELY useful to set your system to stock "Auto" settings as you currently have it configured. Then you can go into the main "Power" menu to review and make note of the Voltage monitor readings, and other features so you know what the ASUS board default settings give you. this then provides baselines for your over-clocking choices.

 

[BonzaiDuck]

For my Q6600, in the submenu for selecting the FSB and RAM/DIMM speed settings, my system FSB is set currently to 1,424 Mhz for the CPU speed of 3.204 Ghz and corresponding CPU_FSB of 356 Mhz. thus, the DDR speed is double that or 712 Mhz, and the system FSB is double that or 1,424.

I've experimented with other settings to get a higher FSB, and was able to get the CPU_FSB up to 375 Mhz with a CAS setting for memory of 4. I believe that if I drop the CPU multiplier all the way to 6, I can go over 400 Mhz -- and one review article on the Striker explains this. I tried it with 7, and I think I was pushing 390.

of course, 7 x 390 = 2.73 Ghz -- not much of an over-clock beyond the stock Q6600 2.4 Ghz speed.

Also, if you lower the multiplier, you absolutely will be unable to ENABLE SpeedStep once you've found stable settings. Apparently, Speedstep depends on the CPU being set to its stock multiplier of 9 (for my Q6600 -- yours will be 10, I think), so that at idle, SpeedStep will drop the multiplier to 6. I think you can see the problem with using SpeedStep at non-stock multipliers -- at least those which are lower than stock.

This by itself is not a major setback, since turning SpeedStep off only disables a power-saving innovation. Most people that I've spoken to, and the poster of our sticky on the Over-Clocking forum, agree that the best performance (not depending on solely synthetic benchmarks) -- derives from a maximum CPU clock using at least the stock multiplier, and a 1:1 divider between CPU and memory speed.

Otherwise, I'd take advantage of my DDR2-1000 potential. Instead, I chose those memory modules with an expectation that the stock (1000 Mhz) latencies fo 5,5,5,15 could be significantly lower at a lower speed frequency, and I was right. I get 3,3,3,6 at DDR2-667, 3,4,4,8 at DDR2-712, and about 4,3,4,8 at DDR2-750.

Your Corsairs, of course, will have a different profile, so I urge you to do some homework and look at forum posts and performance reviews so you don't waste a lot of time experimenting to find optimal latency settings.

 

[rt182]

Mr Yoxxy
so this CPUVVT and FSB termination are those just in the voltages menue =P sorry im alittle abit of a noob when it comes to the voltages and stuff...

Mr bonzai duck right now i have my QX 6700 @ 300 FSB x 10 mulitplier - havent touched my voltages - this is on the gigabyte board BTW i havent installed my second pc with the Striker extreme yet it should be up and running tonight and my Ram @ 1000 mhz (i think that is 500) and 5-4-4-12

i know its not much but its a start im going to try increasing voltage, for my CPU should i increase the voltage on the FSB or CPU Vcore?

 

[BonzaiDuck]

The CPU_VHT boosts the CPU "FSB" voltage. The 1.2VHT is the voltage on the hypertransport link.

You should leave these things at stock values first, and see how far you can push it at a fixed equivalent of the value shown in BIOS voltage monitor under the default "Auto" VCORE setting.

You can look at EVA2000's various settings -- he kept a meticulous log. There was a performance review which I don't have at my fingertips, but the author also gave all the specs for over-clocks at different CPU multiplier settings.

I kept a log of my test-settings over several months -- first for the E6600 and then the Q6600. This is a good idea, so that you remember what the previous settings were when you bump a voltage up and it doesn't help things. Also, BIOS revisions will affect the difference between "set" and monitored values, so it's something you'd best explore on your own with the resources I've mentioned as reference.

This business can be very tedious. The sticky guide on this forum gives good advice, and initially, you may want to try a range of over-clock settings with the same voltages for ten-minutes at a time until you get ORTHOS to fail. Once you have a setting just below the fail-point that is stable, you'd then test it for a couple hours. I finally settled on two or three settings at multiplier 9 and an equal number at multiplier 8, "certifying" them each in ORTHOS runs for between six and 13 hours.

I was more conservative in my VTT, VHT, NB and SB settings than some people recommended here. If a VCORE setting fails, you may then want to bump up VTT by 0.05V, and if that doesn't help, reset that one and try bumping up another voltage by 0.05. And as I said, these experiment logs kept by EVA2000 and featured in reviews (if you can find them on the web) sort of give you some short-cut guidelines.

 

[BonzaiDuck]

Once again -- some of the gurus here on the forums will run their systems with other CPUDR divider ratios than 1:1, and you can do that. I prefer to hold the 1:1 ratios and tighten down the latencies. Fewer clock-cycles means getting the same work done in less time -- or more speed, just as higher megahertz has a similar effect, but there's less stress on the system. I'm not ashamed of my (single VGA-card) 3DMark06 score of 11,600.

 

[rt182]

Originally posted by: BonzaiDuck
Once again -- some of the gurus here on the forums will run their systems with other CPUDR divider ratios than 1:1, and you can do that. I prefer to hold the 1:1 ratios and tighten down the latencies. Fewer clock-cycles means getting the same work done in less time -- or more speed, just as higher megahertz has a similar effect, but there's less stress on the system. I'm not ashamed of my (single VGA-card) 3DMark06 score of 11,600.

what card are you getting 11 600 with

 

[BonzaiDuck]

BFG 8800 GTS 640MB

I used RivaTuner to bump up the GPU core from 550 to 576, and the DDR3 from 792 to 900 (DDR=1800).

I didn't mess with the VGA card's settings until I'd been running my current stable configuration @ (either) 3.0 or 3.2 Ghz for a couple months.

I had memory set to 940 (1880) the other day, and I've tested it to find that it will go to 1030, but I'm "being careful."

 

[BonzaiDuck]

Just some footnotes:

[1] All the 680i boards -- eVGA, nVidia, ASUS, BFG -- have hot-running chipsets. This is due more than anything to the number of PCI_E lanes that are powered -- I think it's something like 49 lanes, 32 of them for SLI systems running each card full-bore at x16.

Some of our water-cooling friends had troubles with 680i boards and switched to the newer Intel chipsets and boards like the P5K -- after pumping up their Q6600's to 3.6+ Ghz. I've speculated that these unhappy experiences arose from pushing the chipset beyond its cooling capacity in WC'd systems that only relied on unenhanced air-cooling for the chipset.

In my motherboard-ducting configuration, I'm using a 120x38mm rear-exhaust fan to suck air off the motherboard (and also from BEHIND it) through a narrow space between a flat duct-panel and the board with its heatpipe-necklace fins and heatsinks. My chipset temperatures, measured by a tape-on thermal sensor, and given an adjustment factor of 3C degrees, is about 41C at idle and 43C at load under ORTHOS-testing and other conditions. [The sensor shows 38C idle and 40C load, plugged into one of the mobo thermal sensor plugs.] These temperatures were taken at a room-ambient condition of just over 70F degrees. The adjustment factor was derived from the discrepancy between another sensor on the GPU heatsink and the GPU temperature reported by the VGA's own internal sensor.

I'd like to think that this "enhanced" cooling regime assists my over-clocking, but others seem to "get there" and farther without it.

[2] Some resellers show the "acceptable" VCORE voltage range as 0.85V to 1.5V. But the Intel "retail-box maximum" for the B3 stepping of the Q6600 is 1.35V. My fixed VCORE setting for my lackluster B3 is 1.32V @ 3 Ghz and 1.41V @ 3.2 Ghz. I can also say this will vary depending on the BIOS version, because later BIOS's were supposed to correct discrepancies between "set" and "monitored." At 3.2 Ghz under load, the monitored value droops to 1.37V. Even the value of 1.41 is less than 5% above the Intel "maximum," and I make a reasonable assumption that they would choose 1.35V as a spec to eliminate under-warranty RMA returns. So there's some probability distributiion of failure rates above that value, and I assume the lower tail of the distribution begins at 1.35. Even our more prudent water-cooling aficianados are setting their VCORE at just over 1.45V for extreme G0 over-clocks as high as 3.7 Ghz. With air-cooling, you're going to increase the temperatures and limit your OC'ing potential anyway if you go that high with the VCORE, and keep in mind that my "ducted" cooling solution makes for a 5 to 8C improvement in TJunction core temperatures over the experiences of people with "good," well-ventilated cases and no ducting.

 

[rt182]

wow you really know your stuff
you wont believe how much this has helped me thanx so much Bonzai Duck
hmmm my striker is up and running currently no OC yet cause i just want to get my second rig functional first if u know what i mean.

now what im going to do with this gigabyte board is prob go return, there is something wrong with the ram slots on my mobo, i get errors when i run memtest but when i put my ram on the striker not one error =S anyway so ill return ask for my money back and buy a 780i board? those are due out soon arnt they

they look pretty good what do u think about them

 

[BonzaiDuck]

I should check at THG and other hardware sites to see what they're saying. This sort of thing creeps up on me, but it was inevitable. Just as soon as several companies had released 680i boards, they began releasing the newer Intel chipsets in mid- to late summer of this year. People jumped ship on the 680i for the P35 and other chipsets. Watch them change again if the 780i is "one-up" on the intel chipset boards.

Pessimist that I am, I'm enthusiastic that the 780i will realize improvements on the 680i chipset, and resolve shortcomings like power-consumption and other factors.

And I'm still wondering whether the Striker will -- or can -- be made Yorkfield quad-core ready with a BIOS fix. Otherwise, your top-of-the-line Conroe/Kentsfield may be the last option left. And of course, under those circumstances, I'll be watching how and when the price comes down on those models as the Penryn Wolfsdale and Yorkfield become an available status-quo.

Unlike some others here, I'm not inclined to jump ship on a technology to catch the very next wave. I proceed at a more modest pace. The last, most advanced system I built before these Striker-based builds was a Prescott 3.4Ghz socket-478.

 

[BonzaiDuck]

I had trouble with a 2x512 Crucial Ballistix DDR2-1000 kit I bought to boost my system from 2 to 3 GB. Since I had the other Striker and E6600 in a test chassis [planning to use it to "play" with VISTA Ultimate], I confirmed that there was nothing wrong with them. I had noticed that they seemed "loose" in their sockets with the first motherboard. I was getting MEMTEST86 errors and failure to complete posting at bootup.

Without doing damage to your motherboard, make sure you push down on the RAM modules so that they're firmly seated. There's enough slop in the latches that you could have a problem like that and mistake it for damaged hardware.

 

[BonzaiDuck]

The ASUS website only confirms Wolfdale-readiness for the Striker. But this Anandtech article suggests that BIOS revision would allow the Striker to apply generally to both Wolfdale and Yorkfield (quad).

Rundown on boards that are Penryn-ready

If that's the case, it will be a while before I swap mobos, although migrating to the 780 chipset may be easy, even if it requires uninstalling the 680i chipset drivers first.

 

[BonzaiDuck]

It appears that the so-called 780i chipset is more like a modified 680i chipset, at
least from a "review" posted on a forum at "bootdaily."

Comment on 780i with close-ups

 

[rt182]

Originally posted by: BonzaiDuck
The ASUS website only confirms Wolfdale-readiness for the Striker. But this Anandtech article suggests that BIOS revision would allow the Striker to apply generally to both Wolfdale and Yorkfield (quad).

Rundown on boards that are Penryn-ready

If that's the case, it will be a while before I swap mobos, although migrating to the 780 chipset may be easy, even if it requires uninstalling the 680i chipset drivers first.

lol the Gigabyte GA-N680SLI DQ6 isnt even on there X P well not rev1 anyway *sigh* this board looks doomed and i think a 780i board should be a nice replacement, hopefully the striker will be nice and compatible

 

[BonzaiDuck]

Unless they were to offer DDR3 and DDR3 was a significant improvement, I think that Penryn-quad compatibility is key, in addition to choice of memory. That's why I'll hold onto these Strikers for a while.

See, people have whined about "FSB holes," "quirkiness," and all sorts of things. I just haven't discovered anything that makes me regret the board. Some user-reviews had people saying that it was "unstable," that their games crashed -- all sorts of things, but I never found it to be true. I gambled that people didn't really know what they're doing.

I figure if I can push a B3 to 3.204, if the memory latencies are lower than many performance DDR_500 kits from the AGP era while running in a range between 667 and 750, it's about as good as it gets without stressing the board. Then, given the memory stock specs, the only thing that's being stressed is the processor, and with these voltages, most of the stress is in CPU_FSB speed. I don't see any "instability," and it looks like a good investment to me. That assessment has been established over 8 months of use.

 

[rt182]

yeh lol hmm just have to wait for now i guess...
give it some time
see what chrismas season brings

 

[BonzaiDuck]

You should probably read this. I assume that you've also at least begun the web-search of other informative reviews -- as I did -- to hit the ground running with this motherboard.


XBit Labs Striker review, February, '07

I'd call the review "positive, but ambivalent." They also seemed tentatively to conclude that there is "variation" in over-clocking potential and performance with different samples shipped and sold, although that observation was made in February, '07, before the release of BIOS version 1004.

All in all, the board has been judged to be too expensive for the variation of experience people have had with it. It didn't work well with certain memory manufactures and models. It DID avoid the problems people had with boards of other manufacture which were pretty much all of nVidia reference design, or, as with the ECS variant, were simply rebadged nVidia reference boards.

So they conclude that it was either THE best, or close to the best of the 680i boards out there. It's overclocking ability was judged less than stellar, but still desirable. And XBit reinforces my own view that you're better off over-clocking at the stock multiplier, and to look for memory bandwidth through tighter latencies as opposed to pushing the FSB to extreme limits.

 

[rt182]

ill definitly go take a look at that, maybe i can more information on overclocking aswel and see what they say

 

[BonzaiDuck]

I was poking around the i4memory.com website and "EVA2000's" posts. They have threads on the Striker -- with updates through the early part of this month, but they're also expanding on threads for boards like the Maximus and the Blitz.

I noticed that EVA2000's recipe of voltages and clock-settings for a Q6600 B3 and the Striker are way out of whack with what produces stability for me. He's using OCZ high-end modules, while I'm using high-end Crucials. He's also using really old BIOS versions, so he's setting his VCORE way up to 1.5 and higher (which would be insane if it didn't produce "real" values like 1.37V or 1.36V).

I strongly urge updating at least to BIOS version 1301, but be careful, don't do it within Windows, use EZ-Flash within the BIOS. Better yet, if you don't mind shelling out, pay $25 to BIOSMan.com and let them do it and send you the PLCC chip and chip-puller. Returning the spare PLCC chip within the next year warrants a free flash of a later version of the BIOS. It's not a bad deal, actually.