Windows Only Sees Sata_0 and Sata_1 on ASR Z97X Killer MB?

[Ken429]

Re: Windows Only Sees Sata_0 and Sata_1 on ASR Z97X Killer MB?

The Mushkin Memory helped the OC situation a little - I was able to get it to run IBT for 10 passes @4.2GHz. However, it needed 1.304V under load and reached 61C. Beyond that it needed way too much Vcore to run @ 4.4GHz - but it did run. I guess that's an improvement since with the GSkill Memory it would not run over 4.0GHz no matter how much Vcore it got. So it's happy at 4.0GHz and runs with 1.216V under load and Core Temps reaching 56C for a moment in time. I'm happy too since this system is much more efficient for a Home Server than the old E8500 or Q9550 systems and seems to be more responsive.

[parsec]

Re: Windows Only Sees Sata_0 and Sata_1 on ASR Z97X Killer MB?

Interesting, I would have been skeptical if I just heard that different memory allowed a better OC, or a lower VCore to get the same speed. But we both are using low voltage DDR3 memory, so a lesson in that, at least maybe for the G3258. Plus think how this goes against the idea of using more voltage when over clocking memory, or a higher voltage from the specs to reach the rated speed. You do see people backing off on their memory speed to keep a CPU OC stable, which may really be doing the same thing, reducing the memory voltage. It makes me wonder if memory voltage commonly limits CPU OCing, or is that the memory itself?

Actually, it seems all the G32xx Pentiums have a rated memory speed of 1333, while the G34xx Pentiums are rated to 1600:



Plus it seems all these Pentiums are compatible with ECC memory.

Really, the G3258 system seems more responsive than a Q9550? I wonder what the IBT GFLOPs of the Q9550 is. I am taken aback by how low the G3258 is, but it is only two cores and threads.

A 4.0GHz OC on a G3258 is a 25% OC of the stock speed of 3.2GHz. How many 4.0GHz processors can OC by 25%, to 5.0GHz? My good G3258 can do 4.5GHz, which is a bit over a 40% OC. I don't know if that is its limit, I'm a little below 1.3V.

But another example of how different these processors can be when they are OC'd. We can see how much leeway there is above the stock speed, which is rather conservative.

[profJim]

Re: Windows Only Sees Sata_0 and Sata_1 on ASR Z97X Killer MB?

Really, the G3258 system seems more responsive than a Q9550? I wonder what the IBT GFLOPs of the Q9550 is. I am taken aback by how low the G3258 is, but it is only two cores and threads.

Stock Q9550 at 2.83GHz LinX results are ~ 38GFlops.

I've found that IBT GFlops scores are slightly lower than LinX GFlops scores.

My Q9550 at 4.20GHz scores:

[Ken429]

Re: Windows Only Sees Sata_0 and Sata_1 on ASR Z97X Killer MB?

I suppose I'm being a unfair when I say the G3258 system is more responsive than the Q9550 system. Especially, when it comes to GFlops. But the Z97 chipset and latest and greatest BIOS versus the P45 chipset and BIOS on the GA-EP45-UD3P MB make a huge difference when it comes to boot times etc. Also the native SATA3 on the Z97 MB makes a difference. I have an add on SATA3 card on the Q9550 system but it is not quite as fast as the Intel Z97 onboard controller. (It's a HighPoint 640L card with 4 lanes so it almost maxes out the current SSD's.) For some reason when I put W8.1 on the Q9550 system and turned it into my Home Server the drives on the HighPoint controller were slow to respond to client requests AND when I went to execute a .exe file it took forever for the program to execute directly from the Server. But that's a problem for another day which hopefully I'll never get to since the G3258/Z97 system with SSD storage makes for a cheap, fast and energy efficient Home Server (Currently the system idles @ ~34W according to the APC UPS). The Q9550 system is back to video crunching which it clearly does better than the G3258.

[parsec]

Re: Windows Only Sees Sata_0 and Sata_1 on ASR Z97X Killer MB?

The results of a default IBT run on my G3258 at 4.5GHz:



The stock speed (3.2GHz) results are ~28GFLOPs (rounding up), so a 40% OC gives me ~5GFLOPs increase.

At a peak of 33GFLOPs, that isn't very high at all. My Haswell Xeon does ~107GFLOPs on the same test, running at 3.8GHz.

I don't get why some people claim the G3258 is good with video games, or I assume it is relatively good in that respect.

[profJim]

Re: Windows Only Sees Sata_0 and Sata_1 on ASR Z97X Killer MB?

The results of a default IBT run on my G3258 at 4.5GHz:

[ATTACH]7675[/ATTACH]

The stock speed (3.2GHz) results are ~28GFLOPs (rounding up), so a 40% OC gives me ~5GFLOPs increase.

At a peak of 33GFLOPs, that isn't very high at all. My Haswell Xeon does ~107GFLOPs on the same test, running at 3.8GHz.

I don't get why some people claim the G3258 is good with video games, or I assume it is relatively good in that respect.

You will get a higher GFlops score with IBT and/or LinX if your test uses more of the available memory, such as 2048MB, 3072MB, etc.
You can artificially increase the score by running the test in windows Safe Mode.

It looks like your G3258 is in the TRTC family: The Runt That Can

[parsec]

Re: Windows Only Sees Sata_0 and Sata_1 on ASR Z97X Killer MB?

You will get a higher GFlops score with IBT and/or LinX if your test uses more of the available memory, such as 2048MB, 3072MB, etc.
You can artificially increase the score by running the test in windows Safe Mode.

It looks like your G3258 is in the TRTC family: The Runt That Can

Yeah I know, I get a bit under 114GFLOPs with maximum memory (out of 16GB) and Windows warns of a memory shortage... Duh!

Using the default settings each of the 10 tests takes ~8.3 seconds, vs the maximum memory test taking 8.25 minutes for each of the 10 tests. I got better things to do...

A runt you say? Albeit accomplished. Plus it was $50 when I bought it, and still is locally. Going by Intel's pricing, 1/4th the price of a Q9550. The Q9550 is seven years old, one step below the top Core 2 Quad processor, the Q9650. That is one small step below, all of 2.83GHz vs 3.0GHz, which is all meaningless once over clocked.

Processors like the Q9550 were overclocked the "old fashioned" way, which is actually the real way it was done before Intel took control of that with the introduction of Sandy Bridge processors. Since then, over clocking has been restricted in many ways, you could even say it is allowed only with certain hardware.

Hmm, MaxMEMM does Ok with Windows 8.1, but has not been updated for a while now. The results seem accurate given the specs:

[parsec]

Re: Windows Only Sees Sata_0 and Sata_1 on ASR Z97X Killer MB?

Proof or it didn't happen:



Show me what an over clocked 4790K can do...

[Ken429]

Re: Windows Only Sees Sata_0 and Sata_1 on ASR Z97X Killer MB?

Ok, I took the bait! What did I do wrong? Why is my LinX results so much different than yours? Also it looks like MaxxMem does not really like the GSkill Ripjaws Z 2400-10 memory.

[parsec]

Re: Windows Only Sees Sata_0 and Sata_1 on ASR Z97X Killer MB?

What did you do wrong? You don't mean that, do you? Seriously...

What I did wrong was use the wrong version of LinX, or really not the same version as you used, and Prof Jim. So...



Now I'm feeling somewhat better, 163 GFLOPs vs 207 GFLOPs, with a difference in speed of 3.8GHz vs 4.6GHz. I can live with that. So while a 40% OC of the G3258 did not buy much, a 20% OC of an i7-4970K (using the 4.0GHz base clock) is worth 44 GFLOPs, although the base clock speeds of these two processors are not identical.

Can't OC an E3 Haswell Xeon, but I knew that, or assumed it would not be possible, and it's true. These E3 Xeons live by the Laws of Turbo 2.0 that 'K' processors ignore. The maximum four core Turbo ratio is 38, two at 40, three at 39. The stock Turbo ratios for an i7-4970K are probably two at 44, three at 43, four at 42.

The next question is what does an i7-4790K do at stock settings? Just wondering...

[profJim]

Re: Windows Only Sees Sata_0 and Sata_1 on ASR Z97X Killer MB?

......
The next question is what does an i7-4790K do at stock settings? Just wondering...

With older systems, the GFLops scores scale closely with cpu speed.
An example:

• Q9550 stock speed is 2.83GHz and the LinX score is ~38 GFlops
• Q9550 at 4.20GHz has a 55.454 GFlops score
• 4.20 divided by 2.83 = 1.48
• 55.454 divided by 38 = 1.46

I'll make a SWAG that this also applies with newer systems too.

[parsec]

Re: Windows Only Sees Sata_0 and Sata_1 on ASR Z97X Killer MB?

With older systems, the GFLops scores scale closely with cpu speed.
An example:

• Q9550 stock speed is 2.83GHz and the LinX score is ~38 GFlops
• Q9550 at 4.20GHz has a 55.454 GFlops score
• 4.20 divided by 2.83 = 1.48
• 55.454 divided by 38 = 1.46

I'll make a SWAG that this also applies with newer systems too.

Ok, running the numbers on the two different (but not very different in base speed) Haswell processors:

4.6 / 3.8 = 1.21

207 / 163 = 1.27

IMO, the result is still close enough to show that correlation.

In your case, the performance gain ratio from your OC is greater than the Haswell processors in this example.

Looking at my G3258 in this way:

4.5 / 3.2 = 1.41

33 / 28 = 1.18

We are actually dealing with percentages here:

The G3258 OC from 3.2 to 4.5 is a 141% over clock.

The GFLOPs performance increase from 28 to 33 is a 118% increase. So the scaling there is poor. That is IBT's GFLOPs, I will try it with LinX.

The same thing applies to all the calculations here. Your 148% OC resulted in a 146% increase in GFLOPs.

The Haswell hybrid has a 121% OC for an increase of 127% in GFLOPs. This is where we must be seeing the result skewed by combining the results of two different processors.

[profJim]

Re: Windows Only Sees Sata_0 and Sata_1 on ASR Z97X Killer MB?

I should have said that the GFlops/speed correlation applies to each processor model and that there probably isn't necessarily a universal constant ratio for all processors.

Another SWAG; the number of cpu cores and amount of L2 or L3 cache probably changes the absolute ratio for each cpu model.


Not to be picky, but your G3258 overclock is 41% higher than stock cpu speed or you could say that your overclocked speed is 141% of the stock speed.
_ A 100% overclock means that your cpu is running at 6.4GHz.

[parsec]

Re: Windows Only Sees Sata_0 and Sata_1 on ASR Z97X Killer MB?

I should have said that the GFlops/speed correlation applies to each processor model and that there probably isn't necessarily a universal constant ratio for all processors.

Another SWAG; the number of cpu cores and amount of L2 or L3 cache probably changes the absolute ratio for each cpu model.


Not to be picky, but your G3258 overclock is 41% higher than stock cpu speed or you could say that your overclocked speed is 141% of the stock speed.
_ A 100% overclock means that your cpu is running at 6.4GHz.

Nope, if I give you 100% of $10, you get $10. 50% of $10 is $5. 150% of $10 is $15.

We commonly use language incorrectly. To say I got a 100% raise in my salary is Ok because no figure or amount was mentioned.

I knew someone my say something about that, which is why I left it there.

[profJim]

Re: Windows Only Sees Sata_0 and Sata_1 on ASR Z97X Killer MB?



Years ago I had a 100% overclock where my E2140 1.60GHz dual core cpu was fully stable at 3.20GHz with 6GB of Ballistix PC2-8500 memory with my P35-DS3L mobo.

Runt cpu's are cheap fun.