- last edited on by
ROGBot
Yippee, yippee! The card is here. Today as nice gift to the Christmass. So before serious review some funny story with screenshots 🙂
Funny story (all this is fictional)
The box from front and back, hmmmm, quickly to the inside
😄 ROG Platinum in blue and white, but....I dont know, this seems not much powerfull. Something strange here.
Of course, maybe join 8 PCIe pin and with more current will be more performance. But still DX9c 🙂
Looking more focused to the box and under protective cover is this with answers:
The card in photos and light description
Backside with massive backplate and voltage points. The side of card show you 2x 8 Power connectors, ROG blinking logo (depends at load of card) and near the SLI is switch for LN2 mode BIOS.
detail of hotwires, LN2 points for soldering and LN2 points (without it doesnt work LN2 mode or memory defrosting)
There are MOLEX for memory defrosting power, defroster jumper and SAFE mode. If your OC corrupt, simply try SAFE mode.
How memory defrosting working in practice? Take a minute for this video:
PS: Now Im preparing for LN2 game event and doing some last test with Rampage review. So be patient with results, next more than 7 days will be Christmas time with bad internet conection (22-29.12.)
Funny story (all this is fictional)
The box from front and back, hmmmm, quickly to the inside
😄 ROG Platinum in blue and white, but....I dont know, this seems not much powerfull. Something strange here.
Of course, maybe join 8 PCIe pin and with more current will be more performance. But still DX9c 🙂
Looking more focused to the box and under protective cover is this with answers:
The card in photos and light description
Backside with massive backplate and voltage points. The side of card show you 2x 8 Power connectors, ROG blinking logo (depends at load of card) and near the SLI is switch for LN2 mode BIOS.
detail of hotwires, LN2 points for soldering and LN2 points (without it doesnt work LN2 mode or memory defrosting)
There are MOLEX for memory defrosting power, defroster jumper and SAFE mode. If your OC corrupt, simply try SAFE mode.
How memory defrosting working in practice? Take a minute for this video:
PS: Now Im preparing for LN2 game event and doing some last test with Rampage review. So be patient with results, next more than 7 days will be Christmas time with bad internet conection (22-29.12.)
Testing Matrix on h2o
Time to show software and first stock benchmarks.
What I must to say, the tmeps in load are simply awesome. Of course, I have better temps than most of people, because I use open benchtable and my ambinets are around 20 C inside....
IInfo about the card+GPU-Z, look at my BIOS version (I do not tested another BIOSes yet) and clocks of GPU card. The basic core clock is 1241 MHz and with Boost jumps to 1342 MHz. But with these temps are clocks still at Boostclocks. Memory working at 7010 MHz effective, type GDDR5 Samsung.
Some time ago I wrote about ASIC, but higher ASIC is not always better. Michal Vobozil said, the best GTX980s are with ASIC 80+% and together with ASIC 67% and lower. My ASIC is 81% and for LN2 OC it could be interesting.... See it later
Asus has GPU Tweak, for Matrix card is this software fully unlocked. We can work with many settings in LN2 card mode. You can set voltages for core, memory, frequency of memory/boostclock, powerlimit, temp limit and PCIe voltage VDD PEX. Even the control frequency controller (VRM).
Settings for GPU-Tweak, bookmark for tweaking
Next settings and hotkeys. You can save your setting into 4 profiles and you can find here 2 predefined profiles also - silent and gaming.
Monitoring in GPU Tweak
First some defaults results
-core i7-5960X@4 GHz and cache 4 GHz also.
-DDR4 memories working at fake XMP profile - 2400 MHz with timings 15-15-16-1T.
-cooling is Noctua NHD14
-my LCD is worst of all components :-), old LCD 24" with FHD resolution, today it could be for 40 dollars 🙂
-nvidia drivers are 344.48
-all tests are without optimizing, this means I do not switch cores, disabled hyperthreading, not core affinity etc
Testy
Oldschool 3D Mark2006, you can download here: http://downloads.guru3d.com/download.php?det=1297
3D Mark11,download link: http://www.guru3d.com/files-details/3dmark-11-basic-edition-download.html
I tested classic Performance test, and in mode Extreme (1080P).
Newest 3D Mark integrated many tests, example good looking Cloud Gate, but its better for comparison older GPU cards or mainstream cards as GTX750Ti/Radeon R7-series or older HD4000/HD5000, latest 3D Mark check here:http://www.guru3d.com/files-details/3dmark-download.html
More usefull is Fire Strike
And harder Fire Strike Extreme (1080P mode)
Unigine benchmarks are popular in last few years. Unigine Heaven fairly tormented your graphics card and partially verify the stability, it is quite long lasting ... Download the latest version here: http://www.guru3d.com/files-details/download-unigine-heaven-benchmark.html
I like more nice looking Unigine Valley, environment of mountains, woods and mountain meadows just did well.
Unigine Valley download here:http://www.guru3d.com/files-details/unigine-valley-benchmark-download.html
As gaming test I tried older DX11 AVP: (http://www.guru3d.com/files-details/aliens-vs-predator-directx-11-benchmark-tool.html) and bechmark based at Crysis engine.
AVP I tested in 1920X1080 with max. details and with all effects active + 4AA/16AF. This means the average 115,1 FPS 🙂
Second gaming test is Monster Hunter.
The world of dragons, monsters, jurassic reptiles and heroic warriors (as me 😛 )
Crysis Engine and added advanced type of rendering, HBAO+ a most of nvidia effects.
Preset test running in 1280x720 resolution
Average FPS are a lot over 100.
At FHD settings are average FPS 71.6 FPS and minimum 40.6 FPS.
What about temperatures??? The card has excellent cooling potential. Spread the paste is great in the foundations, which, paradoxically, we show up in the last part :). Fans are relatively quiet.
Max temps in testing at stock clocks were up to 45-46 C!.
What I must to say, the tmeps in load are simply awesome. Of course, I have better temps than most of people, because I use open benchtable and my ambinets are around 20 C inside....
IInfo about the card+GPU-Z, look at my BIOS version (I do not tested another BIOSes yet) and clocks of GPU card. The basic core clock is 1241 MHz and with Boost jumps to 1342 MHz. But with these temps are clocks still at Boostclocks. Memory working at 7010 MHz effective, type GDDR5 Samsung.
Some time ago I wrote about ASIC, but higher ASIC is not always better. Michal Vobozil said, the best GTX980s are with ASIC 80+% and together with ASIC 67% and lower. My ASIC is 81% and for LN2 OC it could be interesting.... See it later
Asus has GPU Tweak, for Matrix card is this software fully unlocked. We can work with many settings in LN2 card mode. You can set voltages for core, memory, frequency of memory/boostclock, powerlimit, temp limit and PCIe voltage VDD PEX. Even the control frequency controller (VRM).
Settings for GPU-Tweak, bookmark for tweaking
Next settings and hotkeys. You can save your setting into 4 profiles and you can find here 2 predefined profiles also - silent and gaming.
Monitoring in GPU Tweak
First some defaults results
-core i7-5960X@4 GHz and cache 4 GHz also.
-DDR4 memories working at fake XMP profile - 2400 MHz with timings 15-15-16-1T.
-cooling is Noctua NHD14
-my LCD is worst of all components :-), old LCD 24" with FHD resolution, today it could be for 40 dollars 🙂
-nvidia drivers are 344.48
-all tests are without optimizing, this means I do not switch cores, disabled hyperthreading, not core affinity etc
Testy
Oldschool 3D Mark2006, you can download here: http://downloads.guru3d.com/download.php?det=1297
3D Mark11,download link: http://www.guru3d.com/files-details/3dmark-11-basic-edition-download.html
I tested classic Performance test, and in mode Extreme (1080P).
Newest 3D Mark integrated many tests, example good looking Cloud Gate, but its better for comparison older GPU cards or mainstream cards as GTX750Ti/Radeon R7-series or older HD4000/HD5000, latest 3D Mark check here:http://www.guru3d.com/files-details/3dmark-download.html
More usefull is Fire Strike
And harder Fire Strike Extreme (1080P mode)
Unigine benchmarks are popular in last few years. Unigine Heaven fairly tormented your graphics card and partially verify the stability, it is quite long lasting ... Download the latest version here: http://www.guru3d.com/files-details/download-unigine-heaven-benchmark.html
I like more nice looking Unigine Valley, environment of mountains, woods and mountain meadows just did well.
Unigine Valley download here:http://www.guru3d.com/files-details/unigine-valley-benchmark-download.html
As gaming test I tried older DX11 AVP: (http://www.guru3d.com/files-details/aliens-vs-predator-directx-11-benchmark-tool.html) and bechmark based at Crysis engine.
AVP I tested in 1920X1080 with max. details and with all effects active + 4AA/16AF. This means the average 115,1 FPS 🙂
Second gaming test is Monster Hunter.
The world of dragons, monsters, jurassic reptiles and heroic warriors (as me 😛 )
Crysis Engine and added advanced type of rendering, HBAO+ a most of nvidia effects.
Preset test running in 1280x720 resolution
Average FPS are a lot over 100.
At FHD settings are average FPS 71.6 FPS and minimum 40.6 FPS.
What about temperatures??? The card has excellent cooling potential. Spread the paste is great in the foundations, which, paradoxically, we show up in the last part :). Fans are relatively quiet.
Max temps in testing at stock clocks were up to 45-46 C!.
There are temps in killer of GPU, Furmark...At stock values and auto settings for fan regulation are up to 78 C. Take a look for Boost clock, the card has no problems practice boost up to 1405 MHz !
We have here OC part. First with default voltage (VID) and with auto regulation of fans. With this regulation is the card very silent, fans are running around 30-46% in practice (no Furmark). VID of this card is 1.21V, and the core boost is up to 1444 MHz. Samsung memories working at 7590 MHz effective.
OC default voltage, results
3D Mark 11
3D Mark 11 Extreme
3D Mark Sky Diver
3D Mark Fire Strike
3D Mark Fire Strike Extreme
Unigine Valley - HD Extreme
Alien vs Predator FHD, 1920x1080, 4x AA, 16x AF, efekty zaplé
Monster Hunter 1280x720, DX11, 4x MSAA
Monster Hunter 1920x1080, DX11, 4x MSAA, min 42,1FPS, average 74.3 FPS
In Furmark the GPU core holds boost frequency at 1444 MHz and temperatures are around 80 C. We can see the temperature results is a bit higher and power consumption is higher too with stock voltage OC. But about power consumption we will talking more in last part of this minireview. Next two parts will be OC with decent voltage and after overclocking with higher, but still useable voltage settings. Stay tuned :).
OC default voltage, results
3D Mark 11
3D Mark 11 Extreme
3D Mark Sky Diver
3D Mark Fire Strike
3D Mark Fire Strike Extreme
Unigine Valley - HD Extreme
Alien vs Predator FHD, 1920x1080, 4x AA, 16x AF, efekty zaplé
Monster Hunter 1280x720, DX11, 4x MSAA
Monster Hunter 1920x1080, DX11, 4x MSAA, min 42,1FPS, average 74.3 FPS
In Furmark the GPU core holds boost frequency at 1444 MHz and temperatures are around 80 C. We can see the temperature results is a bit higher and power consumption is higher too with stock voltage OC. But about power consumption we will talking more in last part of this minireview. Next two parts will be OC with decent voltage and after overclocking with higher, but still useable voltage settings. Stay tuned :).
After longer time I find time for update. The reasons were my holliday time and after I have not much time for play with new Matrix card 🙂
Max. stable overclocking at air - results[/b]
Currently we are almost at the end, today I will show you maximum stable overclock. Needless to say that the main problem is the temperature and voltage, which cool it is a crazy thing like Furmark test :). Fan speed I had already increased to 100% and temperatures in Furmark initially very surged before they stabilized around 85-87 C. The voltage around 1.325V Im not able cool it in test stability, there is thermal protection card 90 C :).
So final GPU settings:
GPU voltage 1.3 V
GDDR voltage 1.67 V
PEX voltage 1.1V
PowerLimit is set sure to max (but rich enough by around 150%)
LLC 90%
then the match itself and the resulting clocks. The GPU core is finally not so bad - benchmarks handle even a bit higher. By contrast, from memory I waited just a little more. But more about it next time again.
Boost clock GPU core: 1534 MHz
effective frequency of GDDR5 memory: 7937 MHz
What about the benchamrks results after?
3D Mark11 Performance
3D Mark11 Extreme
Unigine Valley na HD Extreme - nicely 85.1 FPS!
3D Mark Sky Diver
3D Mark Fire Strike
3D Mark Fire Strike Extreme
AVP at maximal settings FHD = 129.5 FPS
Lower resolution and Monster Hunters - 132.1 FPS average
At high resolution the game is still quite difficult, though not for the Matrix GTX980 card :). 78 FPS average and minim almost 44 FPS.
Temperatures cards in the tests were more favorable than in a mad bad Furmark:) - around 54 degrees only. Although fanspeeds were high by all and therefore noise cards already worked uncomfortable, but the temperature there was a large margin. In normal operation, I still had the fanspeed set to automatic (but I did not want to reduce them after Furmark test).
Many of us will be interesting about general overview of temperature and consumption reports.
You will see how the card scales with voltage only in terms of frequency and power consumption, but mainly temperature and consumption. I just want to clarify what that means these measurement. The table shows four profiles of frequencies. These I gradually tested in previous contributions to the present one. The first "idle" is clear, I think, after about 10 minutes of assembly at rest in WIndows I wrote consumption from wattmeter. The third column shows the consumption of test "3D Mark 11 Extreme (GPUtest 4)". The following column is consumption in Furmark about 5-10 minutes. Temperatures are then last column and is the temperature in the first case for 3D Mark and the second in Furmark. Besides I have comments from measurements of artifacts, etc., but that's not important 🙂
Max. stable overclocking at air - results[/b]
Currently we are almost at the end, today I will show you maximum stable overclock. Needless to say that the main problem is the temperature and voltage, which cool it is a crazy thing like Furmark test :). Fan speed I had already increased to 100% and temperatures in Furmark initially very surged before they stabilized around 85-87 C. The voltage around 1.325V Im not able cool it in test stability, there is thermal protection card 90 C :).
So final GPU settings:
GPU voltage 1.3 V
GDDR voltage 1.67 V
PEX voltage 1.1V
PowerLimit is set sure to max (but rich enough by around 150%)
LLC 90%
then the match itself and the resulting clocks. The GPU core is finally not so bad - benchmarks handle even a bit higher. By contrast, from memory I waited just a little more. But more about it next time again.
Boost clock GPU core: 1534 MHz
effective frequency of GDDR5 memory: 7937 MHz
What about the benchamrks results after?
3D Mark11 Performance
3D Mark11 Extreme
Unigine Valley na HD Extreme - nicely 85.1 FPS!
3D Mark Sky Diver
3D Mark Fire Strike
3D Mark Fire Strike Extreme
AVP at maximal settings FHD = 129.5 FPS
Lower resolution and Monster Hunters - 132.1 FPS average
At high resolution the game is still quite difficult, though not for the Matrix GTX980 card :). 78 FPS average and minim almost 44 FPS.
Temperatures cards in the tests were more favorable than in a mad bad Furmark:) - around 54 degrees only. Although fanspeeds were high by all and therefore noise cards already worked uncomfortable, but the temperature there was a large margin. In normal operation, I still had the fanspeed set to automatic (but I did not want to reduce them after Furmark test).
Many of us will be interesting about general overview of temperature and consumption reports.
You will see how the card scales with voltage only in terms of frequency and power consumption, but mainly temperature and consumption. I just want to clarify what that means these measurement. The table shows four profiles of frequencies. These I gradually tested in previous contributions to the present one. The first "idle" is clear, I think, after about 10 minutes of assembly at rest in WIndows I wrote consumption from wattmeter. The third column shows the consumption of test "3D Mark 11 Extreme (GPUtest 4)". The following column is consumption in Furmark about 5-10 minutes. Temperatures are then last column and is the temperature in the first case for 3D Mark and the second in Furmark. Besides I have comments from measurements of artifacts, etc., but that's not important 🙂
