Page 3 of 6 FirstFirst 1 2 3 4 5 ... LastLast
Results 21 to 30 of 55
  1. #21
    ROG Guru: Green Belt Array Henkenator68NL PC Specs
    Henkenator68NL PC Specs
    MotherboardRampage V Extreme
    ProcessorIntel Core i7 5960X
    Memory (part number)G.Skill Ripjaws V
    Graphics Card #1R9-290X
    Graphics Card #2R9-290X Matrix Platinum
    Graphics Card #3HD 6970
    Graphics Card #4HD 5970
    MonitorAcer Predator XB720H 144herz 1ms
    Storage #1Samsung EVO 850 1TB
    Storage #2Seagate 2TB
    CPU CoolerCustom Watercooled loop
    CaseCooler Master HAFX and Storm Stryker
    Power SupplySeasonic Gold 1250W
    Keyboard Logitech G19
    Mouse RAT 7
    Headset Logitech G35
    Mouse Pad Steel series
    Headset/Speakers Logitech X 5.1 Surround
    OS Windows 10 Pro 64bit
    Henkenator68NL's Avatar
    Join Date
    Nov 2011
    Reputation
    118
    Posts
    744

    A just one more addition, I always use AIDA64 to get Hardware info, and it has some very interesting information about the GPU:
    - You can see the ATI PowerPlay (BIOS) states. This comes in handy to see where the are at (I have set mine to max).
    - All kinds of other info

    Click image for larger version. 

Name:	2014-03-27_AIDA64_GPU_info.png 
Views:	15 
Size:	112.1 KB 
ID:	35081

    I am currently working on an upgraded version of my Score Calculator and I am integrating a database with hardware / bench scores/ benchtest settings etc. The Hardware info will be pulled from AIDA64 automatically.

  2. #22
    ROG Guru: Diamond Belt Array Zka17's Avatar
    Join Date
    Mar 2012
    Reputation
    199
    Posts
    8,358

    Thank you, Henk, specially for the reference card temp infos!
    Compared to that, my Matrix seems to be much cooler - which is kinda expected with that much engineering in it's cooling system...

  3. #23
    ROG Guru: Brown Belt Array Antronman's Avatar
    Join Date
    Dec 2013
    Reputation
    29
    Posts
    1,374

    So when will you put it on nitro?
    Say hi to the next generation.

    Peace is a lie, there is only Passion
    Through passion, I gain strength
    Through strength, I gain victory
    Through victory, my chains are broken
    The Republic of Gamers shall free me

  4. #24
    ROG Guru: Diamond Belt Array Zka17's Avatar
    Join Date
    Mar 2012
    Reputation
    199
    Posts
    8,358

    Nah, first I want to see what it can do with the original cooling... then water and so on...
    You don't want any hardware to start with more serious cooling than it's original heatsink - changing the heatsink takes time and labor and if the gear is not working...
    Also, you want to know how to OC it first... right?

  5. #25
    ROG Guru: Comunist Array Dartmaul PC Specs
    Dartmaul PC Specs
    MotherboardRIVE
    Processor4930k
    Memory (part number)F3-19200CL9Q-16GBZMD \ F3-17600CL7Q-8GBPIS
    Graphics Card #1R9 290X
    Graphics Card #2R9 290X
    Sound CardXonar DX
    MonitorDell U2913WM + 3 x Asus VW225N (Eyefinity)
    Storage #12 x Samsung 850 Pro 512GB (RAID0)
    Storage #24 x Seagate ST1000DM003 (RAID0)
    CPU CoolerKoolance CPU-370SI LGA2011
    CaseLD PC-V4 Black
    Power SupplyOCZ ZX 1250W
    Keyboard Logitech G11, G13, G710, Steelseries 6Gv2, Func KB-460 (red)
    Mouse Roccat Kone, Kone XTD, Kone XTD Optical, Func MS-3
    Headset Roccat Kave, Func HS-260
    Mouse Pad Steelseries SX. Roccat Raivo, Siru, Apuri, Func F-10L, 1030XL
    Headset/Speakers Logitech Z906
    OS Win 7/8.1/10
    Network RouterTP-Link N600
    Dartmaul's Avatar
    Join Date
    Jan 2012
    Reputation
    67
    Posts
    836

    Z if you can keep 290X under full load below 60 degrees on AIR, it means this certain card has a damn fine cooling.
    I have my ref 290X on H2O and have ~52C GPU temp under contentious load.

    You have made some serious review here, but as an overclocker and 290X owner, I would like to ask you to test few more things those actually matter.
    1) Is this card has a black screen issue. Under high Vcore, most of 290X card are loosing the screen. Monitor reports it has no incoming signal, but whole system (and VGA) are working properly.
    2) Actual VRM abilities. How high would go power draw under high OC, how hot would VRM go in this case, and off course everything about Vdroop.
    3) Memory overvolting affection on mem OC.

    I'm sure you would research these questions sooner or later, don't forget to tell us all about it)

  6. #26
    ROG Guru: Diamond Belt Array Zka17's Avatar
    Join Date
    Mar 2012
    Reputation
    199
    Posts
    8,358

    Thank you, Dart! As I'm reading more and more about the 290Xs, I'm realizing that the cooling on this Matrix is really good...

    No worries, I'm continuing my basic research! Next step will be to see what Power Target does... then Load Line, then the rest...

  7. #27
    ROG Guru: Diamond Belt Array Zka17's Avatar
    Join Date
    Mar 2012
    Reputation
    199
    Posts
    8,358

    It took some time (got busy with work), but finally I'm ready with experimenting the Power Target setting from GPU Tweak!

    The system details and software versions were posted earlier in this thread...

    For this particular experiment I have proposed to change only the Power Target:

    Click image for larger version. 

Name:	PT adjustment only.jpg 
Views:	2 
Size:	583.4 KB 
ID:	35157

    As per my previous testing, the fans were set at max blast for the entire duration of the experiment. I have tried Power Target settings of 100 (default), 110, 120, 130, 140 and 150. And also I have started to OC this card - still at stock voltage settings (1250mV) I have tried to push the GPU Clock by 20MHz steps...

    The ambient temp was 20C during entire time (AC). Every setting was run 3 times, all the items from the Monitor recorded - but only those with the max FurMark scores were used for the comparison. After all we're benchmarking or what... (max scores counts)

    Obviously I've got a lot of data... After careful examination, I have decided to compare only some of them - which I have considered most important of them from the point of Power Target... Please, if you're curious about changes in other parameters, then just let me know! The monitored data I have selected to represent in the charts below are: GPU Votlage, GPU Temp, GPU Clock, GPU Usage, Power Temp.

    Because I have not found a magnifying function in the Monitor, I have extracted the data from the recordings and made charts with higher resolution... I'm showing only the data changes during the benchmark... - time period between the green bars:

    Click image for larger version. 

Name:	GPU Tweak Graphs.jpg 
Views:	0 
Size:	463.9 KB 
ID:	35158

    Again, I have choose FurMark arbitrarily, only because it stresses the graphics card very well and I can set the time of duration... for doing these experiments with Heaven perhaps I would need to take some vacation from work...
    But please, understand that these results are specific for this benchmark, with these settings only! For other benchmarks, if you're going for max scores you have to repeat this experiment by your own... - I'm posting this only to show how the Power Target (PT) setting is working...

    I have wrote the max reached values for the GPU Voltage and GPU Clock. GPU Usage was most of the time at 100% during the entire experiment, it only shows small drops time by time... The GPU Temp was ranging between 31C (idle temp) and max of 67C. Power Temp was starting from 22-23C and most of the time it did not went above 50C, I saw only 2 runs when it maxed at 51C and 53C respectively - but those were not the runs which went bad...

    So then, let's get started!

    My first trial was with default GPU Tweak settings - most important: GPU Clock = 1050MHz / GPU Voltage = 1250mV
    The results with different PT settings:

    Click image for larger version. 

Name:	1050.jpg 
Views:	4 
Size:	328.2 KB 
ID:	35159

    As we can see immediately, just by changing the PT we can end up with pretty substantial score differences... At low PT settings the GPU Clock does not even reaches 1050 which was set... that is perhaps because of the voltage jumps... - practically only the 140% and 150% PT settings gave a stable voltage across the runs... but note that that stable voltage is only 1147mV vs the 1250mV set originally!

    Then I have pushed the GPU: GPU Clock = 1070MHz / GPU Voltage = 1250mV

    Click image for larger version. 

Name:	1070.jpg 
Views:	2 
Size:	334.4 KB 
ID:	35160

    Again, PT 100% and 110% does not gave good results, voltage is all over the place and GPU Clock does not reach the set value... Note that the max score at this speed is not with max PT! The best score came from where the voltage, clock and usage shows a straight line...

    GPU Clock = 1090MHz / GPU Voltage = 1250mV

    Click image for larger version. 

Name:	1090.jpg 
Views:	0 
Size:	339.6 KB 
ID:	35161

    Best score with PT at 130%...

    GPU Clock = 1110MHz / GPU Voltage = 1250mV

    Click image for larger version. 

Name:	1110.jpg 
Views:	1 
Size:	307.9 KB 
ID:	35162

    At this speed I wasn't able to get a straight voltage line, it was peaking to 1060mV all the times... And for some reason the max score did not came from the run with straight clock and usage lines either...

    GPU Clock = 1130MHz / GPU Voltage = 1250mV

    Click image for larger version. 

Name:	1130.jpg 
Views:	0 
Size:	322.5 KB 
ID:	35163

    Interestingly it seems that despite the 1250mV setting the actual voltage feels best on 1147mV...

    GPU Clock = 1150MHz / GPU Voltage = 1250mV

    Click image for larger version. 

Name:	1150.jpg 
Views:	0 
Size:	311.2 KB 
ID:	35164

    It seems that I'm getting close to the limits of the GPU Clock at 1250mV setting... at PT 130% and 150% the screen has started to artifact at around 80% of the benchmark and then: Display Driver Stopped Responding...

    GPU Clock = 1170MHz / GPU Voltage = 1250mV

    Click image for larger version. 

Name:	1170.jpg 
Views:	0 
Size:	219.3 KB 
ID:	35165

    Well, 1170MHz seems definitely too much for the 1250mV setting... low PT setting passed with no problems, but starting with 120% it starts artifacting, then Display Driver Stopped Responding... So, just have stopped right there!

    For going forward, this card clearly needs more juice, but I'm not ready yet for that... first want to see what the other settings does at stock voltage setting!

    SO, the main conclusion from this experiment is that you do have to play with the Power Target! Max settings is not always the best... but most likely you have to reach at least 120% to have the set clocks...

  8. #28
    ROG Guru: Green Belt Array Henkenator68NL PC Specs
    Henkenator68NL PC Specs
    MotherboardRampage V Extreme
    ProcessorIntel Core i7 5960X
    Memory (part number)G.Skill Ripjaws V
    Graphics Card #1R9-290X
    Graphics Card #2R9-290X Matrix Platinum
    Graphics Card #3HD 6970
    Graphics Card #4HD 5970
    MonitorAcer Predator XB720H 144herz 1ms
    Storage #1Samsung EVO 850 1TB
    Storage #2Seagate 2TB
    CPU CoolerCustom Watercooled loop
    CaseCooler Master HAFX and Storm Stryker
    Power SupplySeasonic Gold 1250W
    Keyboard Logitech G19
    Mouse RAT 7
    Headset Logitech G35
    Mouse Pad Steel series
    Headset/Speakers Logitech X 5.1 Surround
    OS Windows 10 Pro 64bit
    Henkenator68NL's Avatar
    Join Date
    Nov 2011
    Reputation
    118
    Posts
    744

    Hey Z, Thanks for sharing. It is definitely in line with the results I have seen. Clear enough is that as soon as you see the core frequency drop, there is a Voltage/power shortage. This is something I have not seen in my there GPU'S, normally when falling short the test/game/driver crashes, but these R series seem to take care of that problem by lowering the frequency. I would love to swap my reference card for that matrix one though!! By the way do you know if the reference water block fit the matrix version? Keep us poster!

  9. #29
    ROG Guru: Diamond Belt Array Zka17's Avatar
    Join Date
    Mar 2012
    Reputation
    199
    Posts
    8,358

    It seems that EK is working on water block for this Matrix card too: http://www.ekwb.com/news/467/19/New-...ks-March-2014/

    Therefore I don't think that the block for reference design will fit...

  10. #30
    ROG Guru: Diamond Belt Array Zka17's Avatar
    Join Date
    Mar 2012
    Reputation
    199
    Posts
    8,358

    As I promised, after the Power Target, here are the results of my experiments with Load Line...

    I was playing with Load Line and I have combined it with the Power Target, the fans were running at full blast - in rest everything else was on default settings:

    Click image for larger version. 

Name:	PT LL adjustment.jpg 
Views:	1 
Size:	588.2 KB 
ID:	35190

    This time I have checked only two GPU frequencies, 1050MHz and 1150MHz. 1050MHz for a base-line and 1150MHz because there I had my first "Display Driver Stopped Running" - it seemed that there is not enough voltage, so theoretically turning off Load Line would provide with some extra juice...

    Here are the raw data - 3 runs, max score values represented:

    Click image for larger version. 

Name:	LLdata.jpg 
Views:	7 
Size:	698.9 KB 
ID:	35191

    Yeap, I had a new issue - the BLACKOUT! At both frequencies with 120% Power Target I have lost the display... I heard about this when reading about 290Xs, this was the first time I have experienced it... After that blackout, I have not pushed the PT more...

    In the table above it can be seen intervals instead of a single value at certain Max parameters... that means that during the run of FurMark the values were not linear, but jumping up and down... - to further investigate the amount of that jumping, I have selected couple of parameters and represented them in higher resolution...

    GPU Clock = 1050MHz (default)

    Click image for larger version. 

Name:	1050.jpg 
Views:	2 
Size:	264.0 KB 
ID:	35192

    GPU Clock = 1150MHz (default)

    Click image for larger version. 

Name:	1150.jpg 
Views:	0 
Size:	265.0 KB 
ID:	35193

    As these graphs show, it really seems that setting Load Line to 0 isn't a good idea... - at least not with this cooling and not with this benchmark... Although the max values of the GPU voltage jumps higher, it can not be maintained, it always results in lower clocks at the same PT settings... With Load Line OFF (set to 0) the GPU voltage goes as high as 1.3V despite it's 1.25V setting!

    So, my conclusion is to be careful with Load Line! This particular benchmark is very very hard on the graphics card, but I can imagine that turning off Load Line would cause similar blackouts with other benchmarks too... - I suspect the temps are going too high...

Page 3 of 6 FirstFirst 1 2 3 4 5 ... LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •