ZOTAC GeForce 9300-ITX WiFi

I wouldn't use the SG05 if you don't plan to be using the included PSU. It's not worth it. You could go with an extremely smaller case considering what you want to put in it.
 
jamesman32 said:
I wouldn't use the SG05 if you don't plan to be using the included PSU. It's not worth it. You could go with an extremely smaller case considering what you want to put in it.
Click to expand...

hmmm...

You're probably right, but I have been looking at cases for months and this one looked interesting. All of the really small cases either don't look quite right, have airflow issues which will make them noisy or are too expensive.

Do you have any suggestions that I might not have considered...

The Origen AE M10 looks nice but its going to be way too expensive, and those 60mm fans can be really noisy...
 
How are quads running on this board? Some users have quads on the board, but how practical is it? Do the quads really heat up the small apex cases that are commonly used here?
About quads, this board only technically supports S quads, the power efficient models. Has any one used the non S quads (95W)? The S quads have a large price premium that I want to avoid. I think that the board may actually support these quads because users have already overclocked their CPUs, meaning more power consumption, possibly on par with the 95W non S quads.

-darkmatter08
 
vertigomhs said:
well most if not all of the c2q's are 95w, and it depends on what case you get, for example the sg05 can fit a gemini ii s which can even let you overclock a little bit.
Click to expand...

I'm sure that would handle a 95W cpu but it's not exactly "low profile" :D It fills up the entire case and cools the entire motherboard. ! You can also get a 300W psu to put in the SG05 so it could certainly handle a quad core as long as it didn't need more than the single 12 volt P4 for cpu power.

There's a thread on the SG05 and the Gemini that shows this setup in case anyone hasn't seen it.
 
Last edited:
i have a sg05 and a geminII s and it fits quite well. I am using it all as a car pc so i didnt need the power supply it came with
 
joao.lnsa said:
Hello, can anyone tell me if i should risk purchasing the Q9400S processor for this board?

It isn´t on the compatibility list but it should be compatible because it´s identical to other processors.

Has anyone tried it? How much of a chance of compatibility do I have?

Otherwise, I will have to go for a much inferior processor, due to the fact that the Q9550S is too expensive at the moment. Will it ever come down in price, like 200 euros or something? That´s the maximum I´m willing to spend on the processor for this system.

Thanks for any help
Click to expand...

I had the same problem and was thinking between Q9550 and Q9550S until I found that the S-Version is NOT worth the premium.

One online review found out (don't have time right now to search again), that you can set the non-S version on par to the S-version by undervolting.
Funny enough, the non-S version could be run stable with a slightly lower voltage than the S-version (!!).
They run both CPU versions and compared performance and power consumption. There are hardly any differences that justify the additional money.
That is no accident, that you cannot find conclusive reports to the advantage of the S-version anywhere on the net.
Because there is none.


Besides this, researching for fundamental differences between Q9550 and E8400 in real world usage, resulted in the fact that a 3.00 GHz dual core does 90% to your complete satisfaction.
The remaining 10% for video processing (rendering & encoding) and GTA IV speaks for 4 cores over 2.
I personally don't like the thought of additional wasting heat and power for a quad if I surf the net, write documents, watch movies, play games, etc etc etc ... like 90 % of the time.
For these kind of things, even the E8400 with speedstep is still overkill.

Whether you have a different user profile than me, you'd have to decide by yourself.
Whatever you do, don't burn your money with the S-version (marketing trick) ....
 
I always expected the processors picked for the "S" versions were basically the same as the non-S versions but tested at a higher quality level. They could be guaranteed to fully function at the S specifications at lower voltages; i.e. slightly better silicon. Lowering the voltage wouldn't be guaranteed to work although it very well might. So Intel won't guarantee a Q9550 would work at the Q9550S specifications but it still might anyway.

Just don't forget to change the bios settings if they get reset. :)
 
Definitely avoid burning money on the "S" versions as the price premium isn't worth it. But do keep in mind that undervolting is not possible on this board either.
 
justinwebb said:
i have a sg05 and a geminII s and it fits quite well. I am using it all as a car pc so i didnt need the power supply it came with
Click to expand...

Are you using a fan on the geminII s? Do you think for the sg05 that a top down cooler is better than a tower push/pull cooler. I just thought that there might be room to fit an 80mm fan on the back with the power supply removed.

The reason I'd like to go for the sg05 is that it has the potential for a very quiet build...

I have had a look at some other itx cases, this one looks promising. Would you be able to keep an E5200 cool in that? There doesn't seem to be much airflow in there...

This case also looks promising, and potentially lots of room for a decent cpu cooler...

I still come back to the sg05, I'd love to put a passive hr-01+ in there, with just the intake fan on the front. Anyone know how much space there is in there without the power supply, has anyone fitted an hr-01+ to the Zotac 9300?
 
Hi Guys, my cpu multiplier seems to be permanently stuck at its default multi in windows.

Im running an E5200 R0 processor with the latest 602 BIOS, my OS is Windows 7 RC.

In the BIOS I have:

Enabled "CPU clock ratio unlock"
Set the "CPU clock ratio" to 8 (I just picked 8 simply to test)
Enabled "CPU N/2 ratio" (this setting seems to enable half multi's)

So in the bios it states the "current valve" as 1700MHz (8.5x200) which to me with the settings ive used makes sense.
Also a clock frequency of 1700MHz (8.5x200) is indicated in the post screen.

1700MHz is just my test frequency so If im running that in windows I know I have successfully manually changed the multiplier.

oh, I have also disabled C1E function in the bios and windows has been set to the "High Performance" profile.

So to my surprise when in windows my E5200 is stated to be running 2500MHz (12.5x200). CPUz also indicates a clock speed of 2.5GHz and I still get the same super PI score as I would at 2.5GHz not 1.7GHz so I know both windows and CPUz are displaying the actual clock speed of the cpu.

All Im after is full control of the multiplier, all looks a ok in the bios but when in windows it almost seems like the bios set multiplier value is over ridden. Im very confused :confused:

Does anyone know If Im missing something very simple :confused:

Cheers :)
 
sheroo said:
Are you using a fan on the geminII s? Do you think for the sg05 that a top down cooler is better than a tower push/pull cooler. I just thought that there might be room to fit an 80mm fan on the back with the power supply removed.

The reason I'd like to go for the sg05 is that it has the potential for a very quiet build...

I have had a look at some other itx cases, this one looks promising. Would you be able to keep an E5200 cool in that? There doesn't seem to be much airflow in there...

This case also looks promising, and potentially lots of room for a decent cpu cooler...

I still come back to the sg05, I'd love to put a passive hr-01+ in there, with just the intake fan on the front. Anyone know how much space there is in there without the power supply, has anyone fitted an hr-01+ to the Zotac 9300?
Click to expand...

yea i am using another fan that pushes much more air then the fan the cooler came with but its still the 120 mm fan. I modded the case so the hard drive would fit as the hard drive cage was touching the fan so i took off the 2.5 in drive cage and just stuck the 3.5 in cage on below the dvd drive to add more room for the cooler
 
for some people like me that had a issue with the e-sata not working on the back panel... to get it to work, youll have to set it AHCI instead of IDE and do a reinstall of the OS.
 
Hi, I just ordered this board for 112.90 Euro = $ 160.12 = £ 97.64
I have ordered a Q9550 (E0) together with the board and 4GB of G.Skill PQ RAM.

Should I do something particular when it arrives ? BIOS update maybe ?

Thx
 
wysiwyg said:
I'm sure that would handle a 95W cpu but it's not exactly "low profile" :D It fills up the entire case and cools the entire motherboard. ! You can also get a 300W psu to put in the SG05 so it could certainly handle a quad core as long as it didn't need more than the single 12 volt P4 for cpu power.

There's a thread on the SG05 and the Gemini that shows this setup in case anyone hasn't seen it.
Click to expand...

gemini II s is a little smaller than the gemini II that you posted. Here's a pic of the profile in my main i7 rig..
 
Hi Chaps, well I gave up with the mysterious stuck multiplier and had a quick go at overclocking the system with the default multi. The ram is far from optimised atm.

Running 0.1v+ over stock voltage and 1.2v NB got me to this...

42ghz1mpi.jpg


I quickly learnt when trying to get my system Prime stable the standard Intel cooler is pretty bad :eek: it quickly became the weakest link! it will run small FFT completely stable but at one points it hit 70c and I bottled it :eek:

405ghzprimesmallfftstab.jpg


I was forced to drop to 0.06v+ and settle for just over 4GHz, still ran prime small fft to hot for my liking though.

This e5200 cpu scales with voltage very well, both the mb and cpu have the potential to achieve a higher stable clock but I will have to pop in a much better cooler to do so!
 
My Configuration :
Zotac 9300 itx

Intel e5200
OCZ GOLD GX XTC 2 X 1 GO PC6400
Western Digital 1T Green
DVD Samsung SH223 
WINDOWS XP SP 3
AHCI enabled in BIOS
APEX MI 008 – 250W
I did not have any problem for about 2 months when suddenly one of the memory stick commited suicide.
Memory has been replaced by others from CORSAIR (667Mhz) till I get my former stick form RMA return.
Since this event, only one core is seen in the operating system and in CPU-Z
I have made several CLEAR CMOS, I took the batterie off.
I have done a BIOS update with the N0422WZT. Nothing new, still the same problem.
Only on core is seen by CPU_Z and the OS.
Is there something I’ve made wrong or something I do not understand ?

Thank’s for your reply
 
wysiwyg said:
Here is the official list. 95W quads are listed but it'll be hard to find a low profile cooler that works with over 65 watts for the mini-ITX cases.

http://www.zotacusa.com/forum/index.php?showtopic=975
Click to expand...

I thought that the board only supported up to 65W processors and concluded the non S versions were incompatible because they were rated at 95W. However because intel is managing to bin enough of the silicon to create an S version, the actual power usage may be lower, even in the non S models. Thanks however.

-darkmatter08
 
darkmatter, you might have read about Intel's Mini iTX board DG45FC that only supports up to the S-Version of Quads (TDP of 65W).

Not sure what you mean if you say that Intel is managing to bin enough silicon to create the S version.
Every single CPU die is been power tested by Intel and put into different frequency classes, like: 2.5 / 2.66 / 2.83 / 3.0 GHz.
They don't bin dies, unless they don't pass.

And as I said before, I read somewhere that a normal Core2Quad had been undervolted and performed even better than the S-Version in regards to power consumption.
 
timme said:
Definitely avoid burning money on the "S" versions as the price premium isn't worth it. But do keep in mind that undervolting is not possible on this board either.
Click to expand...



im really hoping for NEW bios with undervolting options, loosing out on lots og good 12A that could be used on a gpu than producing cpu heat...

i measuret a peak 40W higher with the zotac board than with an undervolted gigabyte X38. thats saying some... :)
 
I did write an email to Zotac Customer Support, asking whether the undervolting feature might be implemented in one of the future BIOS releases.
Their repy was that this is not going to happen, because undervolting could make the system unstable.

Lame excuse.
 
Cov said:
I did write an email to Zotac Customer Support, asking whether the undervolting feature might be implemented in one of the future BIOS releases.
Their repy was that this is not going to happen, because undervolting could make the system unstable.

Lame excuse.
Click to expand...

According to a statement via Twitter, undervolting is not possible on the current PCB.
 
Cov said:
Does that mean undervolting is not a matter of the BIOS but of the PCB ?
Click to expand...
Well, both, actually.

Undervolting in itself is just a way to run stuff (mainly computer hardware, but at least some of those low-consumption light bulbs are also undervolted) below standard operating voltage.

In computers, it of course needs BIOS support, since you need to show the option to the user, and make it selectable. But then you also need component, or generally-speaking, PCB, support. The BIOS code will give out instructions to several components, namely the voltage regulators and clock generator (if I recall correctly). If any of the PCB components can't handle lower-than-standard voltages, you'll either get an instant BSOD (the components don't understand what to make of the order given) or severe instability (in case the components can understand the order and execute it, but can't actually work correctly at such a low voltage).

And, of course, you also have physical PCB limitations. Lower voltages are usually more prone to noise interferences, which means undervolting needs a very clean design. We've already seen in this thread that even 8GB or RAM can be a little too much for this PCB design to handle, so "noise-free" doesn't seem an adjective you can apply to it. Hence the probable reason for Zotac ditching the undervolting option...

Btw, I just love the way Gigabyte usually handles undervolting. If you've seen some of its undervolting-capable boards, you've realized that they don't tell you "well, you have only one voltage calibration option, so that has to work with both load and idle power states", like the other manufacturers do. Which is fine, really, since it's an easy way out. But you have to aim at load voltage values, which means idle states will consume more than actually needed. Gigabyte just tells you "there, you have a standard voltage regulation option", but also asks you "and when the CPU requests the voltage, should I offset it by any value?". That's right, you can just tell it to keep the standard voltage the CPU asks, but reduce it or increase it by a fixed amount to under or overvolt. Nice, huh?

I hope if/when Zotac decides to implement undervolting options, a similar approach is used.

Cheers.

Miguel
 
Cov said:
darkmatter, you might have read about Intel's Mini iTX board DG45FC that only supports up to the S-Version of Quads (TDP of 65W).

Not sure what you mean if you say that Intel is managing to bin enough silicon to create the S version.
Every single CPU die is been power tested by Intel and put into different frequency classes, like: 2.5 / 2.66 / 2.83 / 3.0 GHz.
They don't bin dies, unless they don't pass.

And as I said before, I read somewhere that a normal Core2Quad had been undervolted and performed even better than the S-Version in regards to power consumption.
Click to expand...


What I meant by that is Intel is creating high quality chips that are efficient enough to run at a lower voltage, creating the special S version. Because they are producing so much silicon that is efficient, enough to create an S version quad core, Intel is likely putting the more efficient silicon (the same that is in the S models) into their more popular standard non S Core 2 Quads. That is why I thought that the standard model non S processors were likely using less than 95W.

-darkmatter08
 
darkmatter, I would like to know where you have the information from, that the S-Version uses "more efficient silicon" or are in any other way in higher quality than non-S-Versions CPUs.

I'm particular curious because having read all I could find about the S-Version, did not reveal anything that would confirm your statements.

Thx
 
In any case quoted TDP is not the real TDP it's just a reference number and as seen by silentpcreview it's mostly not worth it. It's not that big of an efficiency hit.
 
I just got a message from Zotac and we have version 01 PCB
PCB 236 DA105 0J1 BF and it seams as tho the identifier is to the right of the"J"
pajopet
 
My board also states "236-DA105-0J1 BF" on the underside.

The board is running fantastically well with my R0 e5200 and its chugging along happily at 4GHz.
I have flashed to the latest 602 bios successfully (even though strictly the latest flash is for v3 boards, but according t zotac it will be fine on any version)

I can now manually change multipliers in the bios but for some reason once in windows it switches back to default multi :confused: Also I still have the S3 standby issue, but I guess that was expected.

pajopet, why are you so concerned with the version of your board??
 
wtf, you flashed to the 602 bios? the note that comes with it says the bios versions are not interchangeable between rev 03 and earlier revs
 
I spoke to Peter Ingram at Zotac and he explained that the 602 bios has been tested on all versions and is fine, basically the bios has advanced further than earlier versions of the board so implementations like dual-link dvi and s3 tweaks are simply going to be ineffective on earlier boards that simply dont have the hardware support.
The 602 bios has better support for R0 stepping processors and from an overclocking point of view seems to be a little more stable than the 422 bios so Im sticking with it.

Edit:

This was the email reply from Zotac when I asked about both R0 support and motherboard & BIOS version comparability.....

Dear Emil,

I contacted Zotac HK about this issue:

"On the Zotac GF9300 ITX WiFi, using current BIOS (2K090602), we have tested the new E5200 (R0 Stepping, Spec number SLB9T) CPU and to perform the best.

Results are positive, CPU name correctly detected, runs fine in Windows and 3DMark Looping test.

I expect the other elements in this "R0 Stepping" E5xxxx CPUs are also supported.

It is also true that the BIOS for the GF9300 move slightly ahead of the hardware, such that the existing BIOS (2K090602) is already added with the codes for "Dual Link DVI". We do tested this BIOS (put to the older PCB which do not support Dual Link), but not found any issue so far."

Kindest regards,
Peter Ingram
Click to expand...
 
Last edited:
Cov said:
darkmatter, I would like to know where you have the information from, that the S-Version uses "more efficient silicon" or are in any other way in higher quality than non-S-Versions CPUs.

I'm particular curious because having read all I could find about the S-Version, did not reveal anything that would confirm your statements.

Thx
Click to expand...

I am saying this because it works in the same way as binning processors post production to create market segmentation - the difference between a Q9550 and a Q9650 for example is just binning. Because of the way processors are produced, they all have flaws in them. Some happen to be better than others and therefore can run at higher speeds or lower voltages - whatever the chip specs are. They are binned into higher line processors. Using our previous examples, the chips that were better and could run within Q9650 specs would be binned and sold as Q9650. The better silicon is more efficient because it may be able to run as a lower voltage than what is specified in the processor's specs because it is a higher quality chip.

Specifically regarding the difference between S and non S versions, the S versions are higher binned chips because they are tested to run at the same speeds as the non S versions with lower voltages. They are more efficient silicon because they successfully manage to run the same speed with lower voltages. That is what I mean by more efficient silicon, and because it is binned higher, it is of higher quality - it performs better.

-darkmatter08
 
darkmatter, you have not answered my question.
Your explanations have no substance and obviously cannot be backed up by any reliable source, therefore I will stop wasting my time.
 
Cov said:
darkmatter, you have not answered my question.
Your explanations have no substance and obviously cannot be backed up by any reliable source, therefore I will stop wasting my time.
Click to expand...
Maybe I can chip in on this one.

OK, so this is how it usually goes in the CPU world: you get a bunch of dies out of a waffer. Each die is individually tested for several parameters, including:
- physical defects (which sometimes ends up producing a smaller-cached CPU line, like the E7xxx and E5xxx series, if part of the cache is bonked; or a single-CPU die, like the 400 series Cellys, if a whole core is not OK);
- maximum achievable speeds;
- maximum achievable speeds at a maximum defined operating voltage.

In the end, you get several different piles of CPUs. Some of them are top-notch, no flaws and great maximum speed, and are put into a special model (XE series, for Intel CPUs; BE for AMD ones).

Others have defects, and can be re-tested as I said above to give birth to other CPU lines. This happens all the time, especially in GPU manufacturing (GPUs are usually bigger than CPUs, and have lower yields, so this kind of technique is very common).

Those who fall in the middle category are labeled according to market demands. Which means there can be great performing parts underutilized because you can't flood the market with too many high-performance parts (too expensive, or it would cause the price to drop, which is not good for profit margins...).

And then there are the "low voltage" parts. Just like the high-performance parts, they are put aside to special models. The only difference is that they can work at the same speed as the regular variants, only at lower voltages, sometime at MUCH lower voltages. It's not that the silicon is more efficient, it's just the etching that worked very well and there are not many current leaks, which makes it work just the same, only needing less power.

And, as with the other standard parts, there are some of these that are "bumped up" to regular parts, because of market demands. So you end up with great low-voltage dies working at higher voltages, just because the market needs it.

Also, it's a known fact that Core-based CPUs can usually work well below standard voltages, so the S series is not that big of a jump(you just pick out a few dies that are guaranteed to work at very low voltages; normal parts may or may not go that low). I wouldn't be surprised if the 9xx0 series had a new stepping in a short while to make it drop in TDP, just like the G0 Q6600 did in its day...

If you need an actual example of how voltage binning works, take a look at the Nehalem parts, both for desktop and server uses. 35xx and 55xx series Nehalem Xeons, as well as Core i7 CPUs come from the very same wafers. The only differences are 1) the socket the dies are mounted on, and 2) the voltage binning. Take a look at the X5570, W3540 and i7 940, for example (same speed, cache size, and QPI bandwidth). LGA-1366 parts have a 130W TDP, the FC-LGA8 part is a 95W CPU. And there are very low-power Nehalem Xeons on the 55xx series, starting from 38W (!) - though admittedly that's a rather slow 2GHz CPU -, with most of them hovering the 60~95W TDP, and only one 130W part. This is all achieved through voltage binning.

As I said, the silicone is the same, only the CPU had a better time engraving, which allows it to work at a lower voltage. In that sense, they are "more efficient", as darkmatter08 put it.

If you need tangible proof, I don't think anyone can give you one. The stuff I just wrote about comes from what i've read about binning (especially GPU binning). Intel is not that fond of letting this kind of trade secrets escape into the wild. But with a high degree of certainty (an "educated guess", as one Autobot once said), that's exactly what happens.

Cheers.

Miguel


P.S.: Sorry for the huge post, and the OT.
 
Thanks for the better explanation. It is what I was trying to explain, but my language could not get the idea across to the readers. I guess 'more efficient' was the confusing part - to me at least it makes sense. If you can do the same amount of processing (work) in less power, then the chip is more efficient. How it does it does not really matter other than the fact that a lower voltage is used to reduce power usage.

Also, has anyone gotten updates on the new revision of the board? Is it being sold anywhere? I know the new one will support wake on USB and 2560x1600 resolution with dual link DVI connections. I have my list of parts ready to order but the only problem is I want the newest revision of the board so I can get 2560x1600 resolution support. Besides those fixes, are there any other differences?

-darkmatter08
 
You must log in or register to reply here.
Back
Top