Hi everyone. Thanks for finding my other blog post. I thought I would come over and say hi and give this post the latest update.
Gigabyte P67A-UD4 Testing
The board that I had a chance to test and run through its paces from the new line of motherboards from Gigabyte was the P67A-UD4. this motherboard has a nice amount of features and a very good motherboard for overclocking.
As usual for a Gigabyte board when you open the box you find a very nice, well done and package with everything you need.
These new black PCB motherboards look very, very nice !
You can find the full released spec's over at Gigabytes website located here.
http://www.gigabyte.com/products/product-page.aspx?pid=3648#ov
The CPU I had to test this with was the 2500K which has 4 cores and no hyper threading.
My very first impressions of overclocking this setup was it had great headroom on clocks with low power usage and the ability to do all this with a good after market air cooler. I used a Prolimatech Mega Shadow with one fan for all my tests.
To begin my testing my goal was to find what my max stable overclock would be, something I could throw everything at and not cause issues.
So lets start off by taking a look at the BIOS screens to see what we have to work with.
I started out with finding my max overclock which when I first started was 4.8ghz, but with a later BIOS update that was increased to 5.4ghz. I then started working that to find stability issues and ended up at a 5.1ghz overclock. This was set in the BIOS at x51 multi and 100 bclck.
CPUz does not show these numbers correctly at this time for some reason.
As we look around we see what different CPU's are clocking at and with what vcores the 2500K that I had seemed to like vcore more than what I see others are using.
RAM speed and timings seemed to be another issue for me. This might have been the kit I using that just did not want to play nice at high speeds on this board but I am sure that with updated BIOS this will be better. The kit I used really did not want to run stable at higher than 1600mhz. I could boot into windows at 1867mhz & 2134mhz and do different things, but some benches would freeze or crash so I went with what was 100% stable for me which was 1600mhz. I could do things like change command rate to 2T and it would be more stable, but the effect of that on benches was rather bad so I tried to find what worked best here and 1600mhz and 1T was it for me.
Another issue I had was mounting heat sinks. The Mega Shadow mounted fine, but the Corsair H70 I have just doesn't line up with the mounting holes and the back plate at all. I am not sure why as it appears that it uses the same pattern as the P55 socket. It's possible that it's my back plate for the H70, I don't know. I did mount the H70 with out the back plate and temps were a little better than the Mega Shadow.
My setup was as follows:
CPU: 2500K
Gigabyte P67a-UD4
Memory: Corsair 2x 2gig sticks, 2000mhz 8-8-8-24 CMG6GX3M3A2000C8
32GB MTRON PRO SSD
Prolimatech Mega Shadow with one fan
Graphics Card: 280gtx
OS: Win 7 64bit stock install
As you can see by my Graphics Card I am some what limited in GPU testing with games and benches. But you know it still does perform pretty well for a older GPU. It's pretty hard for me to keep up with all the latest graphics cards so I tend to be some what behind in those
Finding my max stable OC with these settings.
CPU: x51 multi & 100 bclck
vcore: was 1.600 which is a some what higher than others, but my CPU seemed to like this so all was good.
Lets start off with CPU benches.
Increased RAM speed would help these benches.
I could have run each bench at what ever max speed/RAM setting I could find for each one, but did all on one setting/screen to show stability.
Lets move on to 3DMark Vantage
In all tests that would use Graphics I did overclock the 280gtx to core speed of 725mhz and memory speed of 1200mhz
Cinebench
CrystalMark
Fritz Chess
Nuclearus Multi Core
Performance Test
TMPGEnc Video Encode Test
This used a Rage_Fileshack_Quakecon2007 test file which had a file size of 454,111KB.
Encode time was 1 minute 52 seconds
Well so much for the boring tests LOL
Some of those I included because I have seen others using those benches so I thought I would do the same so you could compare.
How about some game results
I ran the FFXIV bench at both high and low settings. Again this is using the 280gtx so better graphics cards would result in better scores.
The low rez score
And high rez score
Civilization 5
Civ 5 has two different short cuts for starting the game.
One is Civilization 5
Second is Civilization 5 DX11
I do not have a DX11 graphics card at this time, but I ran it both ways.
This is one of my favorite games atm and I usually run it with the DX11 short cut because it does seem to make it look better. The 280gtx handles this game pretty well also.
There is a benchmark for this game but it dumps results into a text file.
Here are my results:
For the regular Civ 5 short cut I get this result
[72230.359] LateGame Full Render Score , 2304, NumDrawCallsPerFrame ,10449
[72230.359] LateGame NoShadow Render Score , 3515, NumDrawCallsPerFrame ,6080
[72230.359] LateGame No Render Score , 19991, NumDrawCallsPerFrame ,0
For the DX11 short cut I get this
[71897.609] LateGame Full Render Score , 976, NumDrawCallsPerFrame ,0
[71897.609] LateGame NoShadow Render Score , 1294, NumDrawCallsPerFrame ,0
[71897.609] LateGame No Render Score , 10149, NumDrawCallsPerFrame ,0
The next benches I ran are more real world stuff which had some very good results.
Cyclone xTreme II Chess Benchmark
Ed Trice's Checkers bench is a single threaded app that really likes pure speed. This one was a very nice surprise and it was the very first setup to break the 3 hour mark.
I have run this bench many times with other setups which included extreme cooling like SS Phase and Cascade. Here we have a $75 air cooler and CPU breaking records compared to the other extreme setups. Very nice !
Other results are:
1 Buckeye Intel Core i5-2500K @ 5.1ghz 02:57:05
2 rickss69 Intel Xeon W3580 @ 4.95 GHz 03:12:40
3 rickss69 Intel Xeon W3580 @ 4.91 GHz 03:14:53
4 rickss69 Intel Core i7-920 Bloomfield @ 4.60 GHz 03:26:05
5 rickss69 Intel Core 2 Duo e8600 @ 4.85 GHz 03:30:25
6 rickss69 Intel Core i7-920 Bloomfield @ 4.50 GHz 03:31:45
7 Kriztoffer Intel Core i7-920 Bloomfield @ 4.39 GHz 03:33:44
8 rickss69 Intel Dual-Core i5-670 Clarkdale @ 4.72 GHz 03:34:47
9 Buckeye Intel Quad-Core i7 870 Lynnfield @ 4.62 GHz 03:35:59
10 Buckeye Intel Core i7-975 Extreme @ 4.32 GHz 03:43:18 Premiere
11 Wei Yang Intel Core i7-870 @ 4.44 GHz 03:45:17
12 Buckeye Intel Core i7-870 @ 4.43 GHz 03:46:00
13 Buckeye AMD 965BE C3 Deneb @ 5.00 GHz 03:46:23
14 Buckeye Intel Core i7-860 Lynnfield @ 4.0 GHz 03:47:16 Mini i7
15 Buckeye AMD X6 1055T Phenom II @ 4.0 GHz 03:51:02 Mini AMD
Another program that is used is Matlab
This program does very heavy duty math related calculations that can take huge amounts of time. It can also generate graphic results so better graphics cards will generate a much better result.
As you can see by looking at the bar graph that this setup performs pretty darn well. In the professional world where time is money this setup can be a real time/money saver.
I have a USB3.0 Buffalo 1TB drive here that I ran through CrystalDiskMark with these results.
This is with the new Turbo USB 3.0 turned on
GIGABYTE On/Off Charge technology
http://www.gigabyte.com/MicroSite/185/on-off-charge.htm
This also worked just fine with my Sprint Samsung Epic 4G phone. It charged while the computer was completely turned off.
GIGABYTE Cloud OC:
GIGABYTE Cloud OC is a free overclocking application that facilitates PC overclocking through any web browsing capable device.
http://www.gigabyte.com/microsite/208/cloud-oc.html
This also worked pretty well. After you install the app you will see this screen.
In the center box you will see a URL to type into the control machine and then get a web page that has OC controls. It actually works very well. I run several headless servers here at home and use RADMIN to remote into these. With Cloud OC installed I can also change overclocks on these machines.
Turbo functions:
I have also read a great deal about the Turbo functions that this platform provides. There is a large number of settings in the BIOS that allows you to adjust Turbo features. This can be very useful to those who want to run a greener system with all the energy saving features turned on.
Normally when us extreme people overclock we turn all these things off so we run at max speed and not have benches be effected by Turbo kicking in and out.
When I setup the BIOS for my tests I also turned off all these features so it should be running at 5.1ghz on the desktop at all times.
I used Tmonitor to check this and found strange results
Tmon is set to show Multiplier changes. I would have expected to see a flat response with Turbo disabled in the BIOS, but instead I see the multi changing as I ran wPrime.
I tested this also on my main rig which is a 975e @ 3.7ghz which also has all these features turned off and see the same thing happening.
So I am not sure if Tmon is reporting things correctly.
Power Usage:
I ran some power usage tests to see how the system performed with that. These tests are the total power usage.
Lets take a look at stock settings with Turbo turned on.
Here we see ~85watts at idle sitting on the desktop.
This had a clock speed of 1.6ghz, mutli set at x16
I then fired up wPrime with 1 thread and you can see the power usage go up to ~110watts, multi jumped to 3.5ghz, a multi of x35.
This is the results of the 3DMark Vantage run @ 5.1ghz.
Idle power is ~138watts with a max power usage of close to 300watts
I had this system crunching for many days in a row and it went with out a hitch. I will have to look at how many points this system cranks out.
You can see that this system at 5.1ghz uses ~225watts for crunching. I would think a person setting a rig up just for crunching would not use a 280gtx but rather something that would use less power. So I would expect that total system power would be lowered with a different GPU.
I did setup this system on my SS Phase unit which ran at ~-40c and saw no real improvement on overclocking abilities, in fact it might have made it worse. So with extreme cooling out of the picture, for now and I do expect that later some one will figure out the tricks or tools needed to give these systems a boost in OC ability with extreme cold. But for now it servers no boost to performance.
To summarize:
This platform is a very good and solid, plus it can serve a wide range of needs, from crunching to gaming and professional needs.
Overclocking is not dead with Sandy Bridge, it's just different, perhaps easier which I feel is a good thing as it will bring more people into the overclocking hobby. It is not just a simple button push to get high speeds, it does take a bit of tuning to find the sweet spots.
Equip your motherboard with a good 3rd party air cooler or water cooler like the Corsair H50 & H70 and you are good to go. I never saw my temps go higher than the low 60c's and in this range these cooling methods hold up very well.
The quality of these new Gigabyte boards is what we have come to expect from Gigabyte, that is very good solid boards with lots of features. The BIOS layout is also something that we have come to know very well as it's pretty much just like all Gigabyte BIOS's with easy to navigate and understand layouts.
Thanks for reading
Gigabyte P67A-UD4 Testing
The board that I had a chance to test and run through its paces from the new line of motherboards from Gigabyte was the P67A-UD4. this motherboard has a nice amount of features and a very good motherboard for overclocking.
As usual for a Gigabyte board when you open the box you find a very nice, well done and package with everything you need.
These new black PCB motherboards look very, very nice !
You can find the full released spec's over at Gigabytes website located here.
http://www.gigabyte.com/products/product-page.aspx?pid=3648#ov
The CPU I had to test this with was the 2500K which has 4 cores and no hyper threading.
My very first impressions of overclocking this setup was it had great headroom on clocks with low power usage and the ability to do all this with a good after market air cooler. I used a Prolimatech Mega Shadow with one fan for all my tests.
To begin my testing my goal was to find what my max stable overclock would be, something I could throw everything at and not cause issues.
So lets start off by taking a look at the BIOS screens to see what we have to work with.
I started out with finding my max overclock which when I first started was 4.8ghz, but with a later BIOS update that was increased to 5.4ghz. I then started working that to find stability issues and ended up at a 5.1ghz overclock. This was set in the BIOS at x51 multi and 100 bclck.
CPUz does not show these numbers correctly at this time for some reason.
As we look around we see what different CPU's are clocking at and with what vcores the 2500K that I had seemed to like vcore more than what I see others are using.
RAM speed and timings seemed to be another issue for me. This might have been the kit I using that just did not want to play nice at high speeds on this board but I am sure that with updated BIOS this will be better. The kit I used really did not want to run stable at higher than 1600mhz. I could boot into windows at 1867mhz & 2134mhz and do different things, but some benches would freeze or crash so I went with what was 100% stable for me which was 1600mhz. I could do things like change command rate to 2T and it would be more stable, but the effect of that on benches was rather bad so I tried to find what worked best here and 1600mhz and 1T was it for me.
Another issue I had was mounting heat sinks. The Mega Shadow mounted fine, but the Corsair H70 I have just doesn't line up with the mounting holes and the back plate at all. I am not sure why as it appears that it uses the same pattern as the P55 socket. It's possible that it's my back plate for the H70, I don't know. I did mount the H70 with out the back plate and temps were a little better than the Mega Shadow.
My setup was as follows:
CPU: 2500K
Gigabyte P67a-UD4
Memory: Corsair 2x 2gig sticks, 2000mhz 8-8-8-24 CMG6GX3M3A2000C8
32GB MTRON PRO SSD
Prolimatech Mega Shadow with one fan
Graphics Card: 280gtx
OS: Win 7 64bit stock install
As you can see by my Graphics Card I am some what limited in GPU testing with games and benches. But you know it still does perform pretty well for a older GPU. It's pretty hard for me to keep up with all the latest graphics cards so I tend to be some what behind in those
Finding my max stable OC with these settings.
CPU: x51 multi & 100 bclck
vcore: was 1.600 which is a some what higher than others, but my CPU seemed to like this so all was good.
Lets start off with CPU benches.
Increased RAM speed would help these benches.
I could have run each bench at what ever max speed/RAM setting I could find for each one, but did all on one setting/screen to show stability.
Lets move on to 3DMark Vantage
In all tests that would use Graphics I did overclock the 280gtx to core speed of 725mhz and memory speed of 1200mhz
Cinebench
CrystalMark
Fritz Chess
Nuclearus Multi Core
Performance Test
TMPGEnc Video Encode Test
This used a Rage_Fileshack_Quakecon2007 test file which had a file size of 454,111KB.
Encode time was 1 minute 52 seconds
Well so much for the boring tests LOL
Some of those I included because I have seen others using those benches so I thought I would do the same so you could compare.
How about some game results
I ran the FFXIV bench at both high and low settings. Again this is using the 280gtx so better graphics cards would result in better scores.
The low rez score
And high rez score
Civilization 5
Civ 5 has two different short cuts for starting the game.
One is Civilization 5
Second is Civilization 5 DX11
I do not have a DX11 graphics card at this time, but I ran it both ways.
This is one of my favorite games atm and I usually run it with the DX11 short cut because it does seem to make it look better. The 280gtx handles this game pretty well also.
There is a benchmark for this game but it dumps results into a text file.
Here are my results:
For the regular Civ 5 short cut I get this result
[72230.359] LateGame Full Render Score , 2304, NumDrawCallsPerFrame ,10449
[72230.359] LateGame NoShadow Render Score , 3515, NumDrawCallsPerFrame ,6080
[72230.359] LateGame No Render Score , 19991, NumDrawCallsPerFrame ,0
For the DX11 short cut I get this
[71897.609] LateGame Full Render Score , 976, NumDrawCallsPerFrame ,0
[71897.609] LateGame NoShadow Render Score , 1294, NumDrawCallsPerFrame ,0
[71897.609] LateGame No Render Score , 10149, NumDrawCallsPerFrame ,0
The next benches I ran are more real world stuff which had some very good results.
Cyclone xTreme II Chess Benchmark
Ed Trice's Checkers bench is a single threaded app that really likes pure speed. This one was a very nice surprise and it was the very first setup to break the 3 hour mark.
I have run this bench many times with other setups which included extreme cooling like SS Phase and Cascade. Here we have a $75 air cooler and CPU breaking records compared to the other extreme setups. Very nice !
Other results are:
1 Buckeye Intel Core i5-2500K @ 5.1ghz 02:57:05
2 rickss69 Intel Xeon W3580 @ 4.95 GHz 03:12:40
3 rickss69 Intel Xeon W3580 @ 4.91 GHz 03:14:53
4 rickss69 Intel Core i7-920 Bloomfield @ 4.60 GHz 03:26:05
5 rickss69 Intel Core 2 Duo e8600 @ 4.85 GHz 03:30:25
6 rickss69 Intel Core i7-920 Bloomfield @ 4.50 GHz 03:31:45
7 Kriztoffer Intel Core i7-920 Bloomfield @ 4.39 GHz 03:33:44
8 rickss69 Intel Dual-Core i5-670 Clarkdale @ 4.72 GHz 03:34:47
9 Buckeye Intel Quad-Core i7 870 Lynnfield @ 4.62 GHz 03:35:59
10 Buckeye Intel Core i7-975 Extreme @ 4.32 GHz 03:43:18 Premiere
11 Wei Yang Intel Core i7-870 @ 4.44 GHz 03:45:17
12 Buckeye Intel Core i7-870 @ 4.43 GHz 03:46:00
13 Buckeye AMD 965BE C3 Deneb @ 5.00 GHz 03:46:23
14 Buckeye Intel Core i7-860 Lynnfield @ 4.0 GHz 03:47:16 Mini i7
15 Buckeye AMD X6 1055T Phenom II @ 4.0 GHz 03:51:02 Mini AMD
Another program that is used is Matlab
This program does very heavy duty math related calculations that can take huge amounts of time. It can also generate graphic results so better graphics cards will generate a much better result.
As you can see by looking at the bar graph that this setup performs pretty darn well. In the professional world where time is money this setup can be a real time/money saver.
I have a USB3.0 Buffalo 1TB drive here that I ran through CrystalDiskMark with these results.
This is with the new Turbo USB 3.0 turned on
GIGABYTE On/Off Charge technology
http://www.gigabyte.com/MicroSite/185/on-off-charge.htm
This also worked just fine with my Sprint Samsung Epic 4G phone. It charged while the computer was completely turned off.
GIGABYTE Cloud OC:
GIGABYTE Cloud OC is a free overclocking application that facilitates PC overclocking through any web browsing capable device.
http://www.gigabyte.com/microsite/208/cloud-oc.html
This also worked pretty well. After you install the app you will see this screen.
In the center box you will see a URL to type into the control machine and then get a web page that has OC controls. It actually works very well. I run several headless servers here at home and use RADMIN to remote into these. With Cloud OC installed I can also change overclocks on these machines.
Turbo functions:
I have also read a great deal about the Turbo functions that this platform provides. There is a large number of settings in the BIOS that allows you to adjust Turbo features. This can be very useful to those who want to run a greener system with all the energy saving features turned on.
Normally when us extreme people overclock we turn all these things off so we run at max speed and not have benches be effected by Turbo kicking in and out.
When I setup the BIOS for my tests I also turned off all these features so it should be running at 5.1ghz on the desktop at all times.
I used Tmonitor to check this and found strange results
Tmon is set to show Multiplier changes. I would have expected to see a flat response with Turbo disabled in the BIOS, but instead I see the multi changing as I ran wPrime.
I tested this also on my main rig which is a 975e @ 3.7ghz which also has all these features turned off and see the same thing happening.
So I am not sure if Tmon is reporting things correctly.
Power Usage:
I ran some power usage tests to see how the system performed with that. These tests are the total power usage.
Lets take a look at stock settings with Turbo turned on.
Here we see ~85watts at idle sitting on the desktop.
This had a clock speed of 1.6ghz, mutli set at x16
I then fired up wPrime with 1 thread and you can see the power usage go up to ~110watts, multi jumped to 3.5ghz, a multi of x35.
This is the results of the 3DMark Vantage run @ 5.1ghz.
Idle power is ~138watts with a max power usage of close to 300watts
I had this system crunching for many days in a row and it went with out a hitch. I will have to look at how many points this system cranks out.
You can see that this system at 5.1ghz uses ~225watts for crunching. I would think a person setting a rig up just for crunching would not use a 280gtx but rather something that would use less power. So I would expect that total system power would be lowered with a different GPU.
I did setup this system on my SS Phase unit which ran at ~-40c and saw no real improvement on overclocking abilities, in fact it might have made it worse. So with extreme cooling out of the picture, for now and I do expect that later some one will figure out the tricks or tools needed to give these systems a boost in OC ability with extreme cold. But for now it servers no boost to performance.
To summarize:
This platform is a very good and solid, plus it can serve a wide range of needs, from crunching to gaming and professional needs.
Overclocking is not dead with Sandy Bridge, it's just different, perhaps easier which I feel is a good thing as it will bring more people into the overclocking hobby. It is not just a simple button push to get high speeds, it does take a bit of tuning to find the sweet spots.
Equip your motherboard with a good 3rd party air cooler or water cooler like the Corsair H50 & H70 and you are good to go. I never saw my temps go higher than the low 60c's and in this range these cooling methods hold up very well.
The quality of these new Gigabyte boards is what we have come to expect from Gigabyte, that is very good solid boards with lots of features. The BIOS layout is also something that we have come to know very well as it's pretty much just like all Gigabyte BIOS's with easy to navigate and understand layouts.
Thanks for reading
Last edited:
