HWBOT Testlab #1: GIGABYTE F2A85X-UP4 P10000+ 3DMark Vantage

  • News, testlab
  • 30
  • HWBOT

0. Introduction

Following the principles described in the About HWBOT Testlab article, we spend some time testing the new FM2-based mainboard from GIGABYTE. The F2A85X-UP4 is GIGABYTE’s most high-end SKU in the FM2 line-up featuring, as the PR reads, technologies such as Ultra Durable 5, Dual UEFI 3D BIOS, Digital Power delivery, triple display support and Lucid Virtu MVP. Over the course of two weeks, we tested the F2A85X-UP4 on several occasions inside the GIGABYTE headquarters.

In the spirit of the HWBOT Testlab philosophy, we conducted our own testing with no interference of any of the vendor representatives, unless requested. In this report, we will provide you as much information as possible on how to achieve the same P10002 benchmark result with the F2A85X-UP4. This includes the BIOS settings, test system details, BIOS files and a full run-down of all preliminary benchmark results. Enjoy!


1. Official Product Description

The flagship GIGABYTE FM2 series motherboard comes in the shape of the GIGABYTE F2A85X-UP4 motherboard, designed to push AMD A-Series APUs to the limit with GIGBAYTE Ultra Durable™ 5 technology. Digital power controllers ensure high capacity, high current power delivery which, when coupled with high speed memory support, combine to push the onboard AMD 7000 series graphics processor to the absolute maximum.

  • Full press release: link
  • GIGABYTE Microsite: link
  • HWBOT Product Page: link


2. Test Configuration Details

To conduct our testing, we used the following components:

  • AMD A10-5800K (x3)
  • GIGABYTE F2A85X-UP4 (PCB v1.0)
  • G.SKILL RipjawsZ F3-19200CL9Q-16GBZMD 2×4GB
  • Corsair AX1200i PSU
  • Der8auer Fusion rev3 CPU pot
  • Windows 7 (64-bit)

The mainboard used was a PVT sample with PCB v1.0. The PVT sample is identical in lay-out and specification to the mass production samples, which means that the results, bios and settings provided in this article are compatible with a retail product. CPU-wise, we had 3 different A10-5800K samples available for our testing, but we only used two (#0003 and #0065). It was the second CPU that achieved the the Testlab Verified P10002.


3. HWBOT Testlab Verified Score

After a two-week testing period, the final Testlab Verified score achieved was P10002 in 3DMark Vantage Performance. The CPU was clocked to 5242MHz, the IMC to 3495 MHz and the system memory to DDR3-2666 CL10-13-13-30. The IGPclk ratio was set to 1275MHz in BIOS which, multiplied by 1.241, results in an approximate clock frequency of 1.58GHz. There is no information on the exact clock frequency as both GPU-Z and the BIOS does not give correct information on current frequency.

Underneath you can find the HWBOT submission information and verification screenshot.


4. Detailed BIOS Configuration Settings

Using Q-Flash, we also saved the used BIOS (including OC profiles) for your information. Disclaimer: using any of the Testlab BIOS or related profiles is entirely on your own risk. Most of the profiles are used for extreme overclocking purposes and are not for regular usage. Don’t try if you don’t know what you’re doing!

For this report, we have two sets of BIOS settings: one for air overclocking and one for LN2 overclocking.

AIR BIOS settings (P8361)

LN2 BIOS settings (P10002)


5. Other Benchmark Results

During our testing we obtained much more results than just the P10002 Testlab Verified one. For the purpose of full transparency, we like to share these initial testing results along with the final score. Hopefuly this will provide you with more information on the scaling as well as give you a couple of targets when doing your own overclocking attempts.

BenchmarkScoreCoolingCPUclkIMCclkDRAMclkBclkIGPratio (IGPclk)HWBOT linkScreenshot
3DMark11 Performance2879LN2499234941333124.81275linklink
3DMark11 Performance2788LN2499334941331124.81175N/Alink
3DMark11 Performance2672LN2511130141223131.11150N/Alink
3DMark11 Performance2511LN2511130141048131.11150N/Alink
3DMark11 Performance2500LN2419430141048131.11150N/Alink
3DMark11 Performance2391AIR432522281223131.11175N/Alink
3DMark11 Performance2372LN2314526211048131.11100N/Alink
3DMark11 Performance (GT1)10.89LN2511734941331124.81275N/Alink
3DMark11 Performance (GT1)10.45LN2499334941331124.81175N/Alink
3DMark11 Performance (GT1)9.98LN2511132761223131.11150N/Alink
3DMark11 Performance (Physics)20.09LN2599234941331124.8N/AN/Alink
3DMark11 Performance (Combined)9.67LN2511132761223131.11150N/Alink
3DMark Vantage Performance10002LN2499334941333124.81275linklink
3DMark Vantage Performance9945LN2499334941333124.81275N/Alink
3DMark Vantage Performance9754LN2499334941333124.81175N/Alink
3DMark Vantage Performance9321LN2511134071223131.11150N/Alink
3DMark Vantage Performance8548LN2419430511048131.11150N/Alink
3DMark Vantage Performance8361AIR445623591223131.1925N/Alink
Memory Clock1398LN2N/A34081398N/AN/Alinklink


6. Media Gallery

7. More Information

For more information on the HWBOT Testlab, please check our introduction page: link.

If you have any questions or concerns regarding this or any other HWBOT Testlab report, feel free to get in touch through email or the forums. We are always interested in hearing constructive criticism and new ideas to improve the validity of the Testlab results.


30

United States xxbassplayerxx says:

I thought these chips were hitting 7GHz+ on LN2. Is it because you were using the IGP that it didn't go any higher?

Belgium Massman says:

Yes. The board is limited to 1.80V Vcore, if I remember correctly, and I was hitting about 6.5GHz on both chips with that voltage. The biggest problem is that the APU will shutdown if you use too much voltage/frequency for either the CPU or the IGP. So, in order not to trigger the shutdown I kept the CPU at relatively low voltage and frequency. Besides, the effect of CPU frequency is almost insignificant anyway. 1GHz additional CPU frequency equals a total of 11pts boost in 3DMark11.

United States xxbassplayerxx says:

Massman said: Yes.

The board is limited to 1.80V Vcore, if I remember correctly, and I was hitting about 6.5GHz on both chips with that voltage. The biggest problem is that the APU will shutdown if you use too much voltage/frequency for either the CPU or the IGP. So, in order not to trigger the shutdown I kept the CPU at relatively low voltage and frequency.

Besides, the effect of CPU frequency is almost insignificant anyway. 1GHz additional CPU frequency equals a total of 11pts boost in 3DMark11.


Wow... I thought the Physics portion of the benchmark had a pretty large effect. Guess not!

Australia Dinos22 says:

cool report man

Belgium Massman says:

xxbassplayerxx said: Wow... I thought the Physics portion of the benchmark had a pretty large effect. Guess not!


It's a balance-thing. For the Physics score to have a large effect, you also need your GPU score to be high.

Germany der8auer says:

Nice report pieter!

Czech Republic froxic says:

Very nice, settings will very be usefull for my next LN2 session! Btw in F3E BIOS can be CPU voltage adjusted to 2,2V ;-)

United States sin0822 says:

yea the new one has more vcore option, nice job pieter

Brazil Raul SDT says:

http://www.criminalcafe.com/showthread.php?10155-Killer-Trinity-top-A10-5800k-confirmado-5.1GHz-no-air-e-7.3GHz-no-LN2

You guys should take a look at this ^^
:)

Russian Federation DKZ says:

AHCI MODE? OR IDE MODE

Belgium Massman says:

Ide

Indonesia [K]vlolv says:

Massman said: The biggest problem is that the APU will shutdown if you use too much voltage/frequency for either the CPU or the IGP.


Wondering this a new bugs like llano when used D-Sub such have a limiter or maybe basic trinity characteristic

San Marino jp7 says:

is there a way to disable OCP with newer BIOS? i have strange issue over 1.6V with 3dmark vantage

Belgium Massman says:

What is the strange issue?

San Marino jp7 says:

Massman said: What is the strange issue?


changing from 4.7GHz 1.5V to 4.7GHz 1.6V get half OPS in CPU test 1 and 2

Belgium Massman says:

If the problem would be OCP, the system would just shutdown. This sounds like throttling in some way. How are the temperatures?

Finland The Stilt says:

jp7 said: changing from 4.7GHz 1.5V to 4.7GHz 1.6V get half OPS in CPU test 1 and 2


Or air cooling?
If so it is most likely either VR_Hot (VRM) or CPU HTC activation (@ 70,0°C).

Enable "PowerMode" with TCI K2 and see if it makes any difference.

However it certainly isn´t VRM OCP.

Germany der8auer says:

I don't understand why I can boot up to 145 BCLK but just 106 into windows? And only 102 BCLK are benchstable... Also I can not go above the 1866 memory divider. If I used 2133 or above it doesn't boot anymore. Pretty much same settings / voltages like Pieter used in the first post.

Belgium Massman says:

For high FSB clocks, you should try the 'even' memory multipliers: either DDR3-800, DDR3-1600 or DDR3-2400. Also set IDE, not AHCI.

Germany der8auer says:

Thanks PJ. Gonna test it now.

Belgium Massman says:

Fyi, around 133MHz BCLK (forgot whether it's 133 or 140) the board will switch to the external PLL. So, you might want to go in small steps there. Something like
[list]
105 -> 130
145 -> 170[/list]

Around 170MHz -175 to be precise- was the highest the board could do with multiple CPUs. Afaik, it was related to the used clockgen (you could do a little higher fiddling with clock spectrum).

The boot-in-windows issue is 99% IDE/AHCI related.

Germany der8auer says:

Yes it was IDE related :) The memory issue was due to wrong subtimings. I had to set all the subtimings manually. Now I can also use 2133 MHz divider. I'm aiming for about 120-125 MHz BCLK for benching. According to Hicookie the board switches to the external PLL at 133 MHz like you said. Thanks!

Germany Moose83 says:

Now go Roman, and crush the Trinity for Germany :D :p

Germany der8auer says:

Okay I'm realy annoyed now.. Spent 3 days to get the correct performance and to unbug the system with these crappy AMD drivers and when I just wanted to start the 2nd Lightning died out of nowhere on stock clocks.

San Marino jp7 says:

The Stilt said: Or air cooling?
If so it is most likely either VR_Hot (VRM) or CPU HTC activation (@ 70,0°C).

Enable "PowerMode" with TCI K2 and see if it makes any difference.

However it certainly isn´t VRM OCP.



SS cooling, is not related to temperatures, CPU\GPU under zero and VRM are not hot, after some run 22/24° ( checked with TCI K2 )

enabled PowerMode but doesn't change the situation.

it is very strange... I can't understand why there is this sort of throttling however tomorrow I am going to format :) and reinstall OS but I am not confident :(

Belgium leeghoofd says:

Where can we find that S1 bios ? need to improve my pi time drastically :)

Belgium leeghoofd says:

thx you are so close to being an Angel roman :p

Belgium Massman says:

Leeghoofd said: Where can we find that S1 bios ? need to improve my pi time drastically :)


Under section 4 of the article ... ;)

Belgium leeghoofd says:

Too late used F3K yesterday for my last contribution of the year

Please log in or register to comment.