My Dual Harper - Asus Z7S WS living review *56K warning*

[Jaco]

Hi Joschi,

My Asus Z7S is up & running . No problem at all, except the ram issue.
I'm going to start overclocking now

[jcool]

Great to hear, let me know how it goes. You got 2 CPUs yet? Also E5420's?
About the Ram, you'll figure it out. It can't be so hard removing these damn heatspreads, can it?

By the way, is the "package" under way yet?

[MaxCFM]

Date/Time Activity
06/04/2008 DELIVERED[D] MB INTEL BOXD5400XS 771 5400 RT

Maybe I can catch you and Dave now in hwbot If nothing else I'll add a new cruncher!

[Jaco]

Great to hear, let me know how it goes. You got 2 CPUs yet? Also E5420's?
About the Ram, you'll figure it out. It can't be so hard removing these damn heatspreads, can it?

By the way, is the "package" under way yet?

1 CPU for starters , E5420 . Hacking the heatspreaders is for later , I'll find a way . definitely glue involved , it doesn't move at all.
Package is for tomorrow

[ZIO]

I watched my temps drop over the first 24 hours, they stabilized after dropping to 33c after all day use by 2 heavy users.

Idle temps:
cpu0 33c/91f (pretty cool compared to the 105F earlier in the day.
cpu1 33c/91c

mb 32c/89f

after 16.35 minute load test = 35c/100F (ripped a 1.5hr DVD into 30fps ipod video in 16.35 mintes)

Then right back down to cpu 33c/91F

Current Build:

MB Asus Z7S
2, 2.5Ghz Harpertown CPU's
Spinpoint 1TB SATA HD
2GB RAM with MASSCOOL heatpiped copper sinks (more on the way)
2, HIS 512MB Radeon 2600XT Ice-Q Turbo, (crossfired)
22" LG 3000:1, 2ms crispy LCD
Antec Sx835ii, modified Server Case, 5 case cooling fans

Both CPU's are idle at same temp, when lightly loaded increase to 35c/100F

I'm gonna wait a week or two before OC'in these let em burn in good 24-7.

Other than that IT's ALIVE!!

This build is the Shiznit. After all the hoopla, it's reads right, whoah fast.

Thanks jcool.

[Jaco]

Congrats Zio ! Good luck !

I'm at 3.0Ghz (400x7.5 ) since yesterday. No way i could wait one whole week to start overclocking

[jcool]

Date/Time Activity
06/04/2008 DELIVERED[D] MB INTEL BOXD5400XS 771 5400 RT

Maybe I can catch you and Dave now in hwbot If nothing else I'll add a new cruncher!

Depends on the CPU's you're getting. Max FSB sucks with Skulltrail though...

Thanks jcool.

No problem
Seems like I started a trend buying this mobo
Asus should send me another one for all the advertising I've done

[WesM63]

Date/Time Activity
06/04/2008 DELIVERED[D] MB INTEL BOXD5400XS 771 5400 RT

Maybe I can catch you and Dave now in hwbot If nothing else I'll add a new cruncher!

Yep, you're going to need some hefty cpu's to catch dave. That or some very good FB-Dimm's.

[jcool]

Ok guys,

as you know, I have installed another 2 rams to enable QuadChannel, but couldn't prove until now that it was really active. No increase in crunching or Cinebench, CPU-Z doesn't read it, it doesn't show in Sisoft Sandra etc.

After I just looked at the Nehalem@Anandtech article it came to me... Winrar! Best program when it comes to measuring mem bandwidth.
So I paused BOINC for a second, fired up winrar, hit alt + b and this is what I got:





This simply blows my 4Ghz, DDR2-1100 X38 mainrig away. Guess it's really working, not bad for DDR2-860 5-5-5 eh

[CyberDruid]

Wow you went through a lot to get that all dialed in. Good Job.

[[XC] NetburstXE]

Those 8 little graphs are sooo hot

[ZIO]

Hey peoples-

I think we should compare FSB hot spots for other OC'ers.

Like this:

Harpertown 2.5Ghz OC Heat chart

Idle temps:
CPU:2.75Ghz = 95/96F, 366FSB
CPU:2.80Ghz = 98/102F, 380FSB
CPU:2.92Ghz = 91/93F, 390FSB

I found a hotspot at 380FSB, and then I found a COOL spot at 390FSB, I mean the processer is running at the same temps at 2.5Ghz!!

This indicates there are going to be regular cool spots thoughout the Processor FSB, and mapping these could lead to superclocking maps for this series of processor, instead of dialing up through hot zones, you could ballpark and tune up (OC) through those cool spots without fracturing the circuit pathways, ie: blowing up your chip by tuning through hot FSB zones.

I think that if done correctly we could get a super-clocking chart made for the harpertown that might even allow 4Ghz on this build.

Btw if my MB says my RAM is 667Mhz, then is it not 800? or do I have a setting or something wrong somewheres? If it's 667Mhz, then I need to call my supplier.

Thanks in advance-

ZIO

[STEvil]

What ram did you buy?

[kevinbo03]

So i skimmed through this whole thread, But i didn't see one thing i was curious about... How many ppd is this beast putting out?

[Martijn]

So i skimmed through this whole thread, But i didn't see one thing i was curious about... How many ppd is this beast putting out?

Someone correct me if I'm wrong, but IIRC these systems should do ~35k/day.

[ZIO]

What ram did you buy?

Kingston valueram.

KVR800D2D8F5K2/2G

[jcool]

Btw if my MB says my RAM is 667Mhz, then is it not 800? or do I have a setting or something wrong somewheres? If it's 667Mhz, then I need to call my supplier.

Thanks in advance-

ZIO

The mobo says 667 for the mem because it detects FSB 333 CPUs, and the Z7S always runs the mems with a 1:1 divider (at least if you use DDR2-800 rams).

So If you're running the CPUs at FSB 390, your mem runs DDR2-780, just like mine runs 860 at FSB 430.

Dunno about the hot spot - cool spot stuff, I just went as high as they were willing to go FSB-wise.
My experience is that they don't need to be cold (mine are running fine between 60 and 67C realtemp).

@kevinbo03: I don't know really, all I can say is it rocks

[ZIO]

Dunno about the hot spot - cool spot stuff, I just went as high as they were willing to go FSB-wise.
My experience is that they don't need to be cold (mine are running fine between 60 and 67C realtemp).

jcool,

I'm just wondering if the harpertown has a FSB cool slot somewhere up higher, like 3.7 or 3.9Ghz maybe it would run at 67c? If you dialed up through the cool zones (one of which should be 390fsb), the processor is likely to have cool zones from synced frequencies overlayed on the 45nm structure, since heat is based upon resistance which is a form of vibration found within circuit pathways and the gates themselves. Put simply, certain frequencies will produce less heat due to frequency interactions with the transistor gates.

There may be a way to go even higher if dialed correctly. If the chip is stable at full load at 67c then I think this approach is possible. Making 67c the high end and paying attention to temp while dialing.

I'm thinking we could go even higher than 3.2Ghz if we knew which FSB frequencies produced the least amount of heat. So a map of cool zones is needed, from which a pattern could be found. Maybe making a cool zone map from 2.5Ghz to 3.2 would reveal it?

More than likely I think that actual testing will reveal the map, someone would actually have to try to dial up their FSB and record the temp responses.

Because we are dealing with a crystal it has very specific vibrational characteristics, at certain "right" frequencies it will not generate heat as much. However at the wrong frequencies it will light up like a heating element.

I'd love to see a 4+Ghz 8-core machine!!

A bit long-winded but a good cause to be sure, understanding heat mechanics in molecular solids and machines is one of my understandings from quantum vibrational mechanics.

ZIO

[jcool]

Very interesting stuff there. Hate to break it to you, but FSB 430 isn't just the CPU's limit. The mobo is under heavy strain there, especially since I'm running 8GB ram and Quadchannel. It might handle 440 FSB 24/7 if the CPUs do their part, but that's about it. No FSB500 happening on a i5400 chipset, even 400+ will remain a dream for most.

Also temps aren't everything. I see that every day with my Q6600, which runs at 4Ghz at -39C evap temp just as fine as it runs under water with 65C Coretemp at the same frequency and voltage. Only I can't keep it that cool for long with watercooling, hence the phase change.

4Ghz+ octacores are very much attainable, if you have the money. Just need some QX9775's, good cooling and you're set

Also no cold spot it going to make a quad run FSB500, let alone two of them and never ever w/o detailed VTT and GTL adjustments that only very few motherboards provide.

I'm afraid I can't help you there, but maybe Dave can, as soon as he's got his E5450's. They have the kind of multiplier you need to get close to 3,8Ghz. FSB-wise it's not happening to the smaller ones, even if you hit the mother of all cool spots.

[jcool]

Hi Fellas,

today a new bios for the Z7S WS appeared on the ASUS ftp. Version is 0302, no changelog yet but I'll give it a try and let you guys know

Download it here

[Jaco]

New bios ,that's good news. thanks !
***hopes for vcore adjustments cuz my E5420 needs more juice ***

I'll try it later ,hopefully tonight , no time right now.

[Jaco]

New bios is installed on my Asus Z7S.
No new options in the bios at first sight.

First impressions :

CpuZ opens up normal (took 4-5 min with the 0201 bios)
Small performance increase : judging by some quick Superpi runs.
Looks fine to me as far as i can tell.

E5420 2.5Ghz overclocked 430 x 7 = 3010Mhz
With the old bios (0201):

superpi 1M 16.344
superpi 2M 42.235
superpi 16M 8m57.641

E5420 2.5Ghz overclocked 430 x 7 =3010Mhz
New bios (0302)

superpi 1M 16.156
superpi 2M 41.781
superpi 16M 8m54.969

[redmodel]

Finally put mine together. A few issues.

I am having the exact same issue as I did with the DSEB-DG board. Testing with two different brands of 667mhz FB-DIMMs, Prime95 will crash using high FFTs (above 2500). But it didn't do it with one piece of MacPro 800mhz ram by Hynix.

Does anyone else mind testing this? I'm interested in knowing if people with 667mhz ram is getting the same issue, and if people with 800mhz is not.

I don't understand why it's doing it, i am use e5450 cpus, which should be using the 667mhz ram anyways. but I want to overclock it to 3.6ghz, and I planned on getting 800mhz ram either way.

Other issue is with configuring the Marvell Network drivers to do link agregation. That is what was nice about the DSEB-DG. It had 4x intel Pro Network Ports. Zero issues. The Marvell works fine as a single link. But trying to team them up has been causing a lot of problems, to a point where I can't even get a proper connection anymore, even on a single link. I don't have any slots left either to get an intel network card

Third Issue, my eSATA drives freezes up the computer whenever I have it plugged in to the Jmicron port. I remember seeing this before on another build, and at the moment I don't remember how I fixed it. I have to look into it more. Just wondering if anyone has noticed this?

The other issue is changing the SATA IDE to AHCI drivers in Windows after install. This trick used to be easy on other boards, but I get iastor.sys crashes now whenever I do it. oh well.

[neveroffline]

Hi Guys.

I finally got my new system (thanks to jcool for helping me with this):

Apevia Jupiter X Full Tower
Corsair 620 HX PSU
Vista 64 Bit Ultimate
Asus Z7S (rev 0201)
2x Diamond Viper 3870 HD 512MB GDDR4
Transcend 4x2GB DDR2 667Mhz FBDIMM (8GB) with ram cooler (dual fans)
2x Seagate Barracuda 250GB Raid 1
2x DVD Burners w/LightScribe
2x XEON 5410 2.33Ghz Harpertowns (stock cooling)
5x 120mm Fans (2 in, 3 out)

Displays: 32" LCD 1080i, 24" LCD 1080p, 21" LCD 720p

I actually got this all from newegg last week but had some issues. The board did come with the 0201 bios, so didn't need to flash it. Had everything together, ran overnight, in the case, on it's side. Next day went to move it vertical and it kept shorting/grounding/not posting which I couldn't fix as at the time I was sure it was a mb failure. Sent that one back and now I'm waiting with electrical tape in hand for each standoff and the closest solder points on the back to attempt to prevent any grounding on vertical in this case.

Also unfortunately the case I got has the cpu standoffs for a full size EATX, and not the Z7S, so I ended up using the supplied backplates with the 6L standoffs. I must've mounted the board a dozen times trying to prevent bending and the short on vertical - eventually I gave up and decided I wanted to RMA it and start from scratch with the board again.

For the time it did run my temps were about 95-110F for both cpus, and about 105-113F for the ram. I didn't do any overclocking, and simply left it in the bios overnight.

I'd like to get my 5410's up to hopefully 2.6, or 2.8 but would enjoy anyones help.

Also a couple questions, is anyone using the backplates in the case or do you all have the proper standoffs? The suckers are stuck on my case very well, and seem to provide ample support when the heatsinks were attached, but the mb did have some bending. Also did anyone else experience any grounding/shorts when placed in the case and vertical? I was thinking of using nylon standoffs, but asus said to simply cover the brass with an X of electrical tape. I'm open to your suggestions here as well.

Special thanks to jcool who helped me through email pick all these parts.

I should have the new Z7S this Monday, so will be ready to jump back into it then.

Dust

[jcool]

New bios is installed on my Asus Z7S.
No new options in the bios at first sight.

First impressions :

CpuZ opens up normal (took 4-5 min with the 0201 bios)
Small performance increase : judging by some quick Superpi runs.
Looks fine to me as far as i can tell.

E5420 2.5Ghz overclocked 430 x 7 = 3010Mhz
With the old bios (0201):

superpi 1M 16.344
superpi 2M 42.235
superpi 16M 8m57.641

E5420 2.5Ghz overclocked 430 x 7 =3010Mhz
New bios (0302)

superpi 1M 16.156
superpi 2M 41.781
superpi 16M 8m54.969

Yeah I think Cinebench improved a bit, too.
Other than that, there aren't any major changes, what I noticed:

- FSB spread spectrum option is gone, had it disabled anyways
- they added something to the SATA menu, like compatible or enhanced mode for IDE... don't really get it so I left it alone (there still is the main Sata mode option with enhanced, compatible, IDE; AHCI, Raid etc)

@redmodel: I don't know about 667 mems, but I can assure you I am 24h prime blend stable at FSB 430 with mem running synced. However, a friend of mine noticed strange behaviour with 667 ram when running a FSB 333 CPU. His measured ram bandwidth was extremely low, especially when he overclocked the system to FSB400. Didn't have that problem with 800 rams, and we're all using them around here, hence no problems. The Z7S doesn't seem to like 667 mem.
Dunno about LAN teaming, I'm hard-pressed to get my gigabit network to perform (always stuck between 15 and 22% grrr). Oh and I don't use eSATA, can't help you there either.

@Dustin:

Glad you finally managed to register with XS forums
I don't really see a problem with shortening the mobo, I ran it for like 3 weeks on a table and I use stadnard metal standoffs in the case, no problems whatsoever.
I do recommend using a case featuring the mountings for the CPU coolers, otherwise you'll be stuck with a bending mobo, especially if you use HR-01 X heatsinks.

About the temps, I'd recommend replacing the NB sink with a Thermalright HR-05 IFX or, if you can get it, the Alpenföhn Ötzi (has a bigger base plate).
You'll have to actively cool the mems too, unless you mount special heatsinks on them.