Overclock.net › Forums › Industry News › Hardware News › [AdoredTV] Pascal vs Maxwell at same clockspeeds, same FLOPS (1080 vs 980 Ti)
New Posts  All Forums:Forum Nav:

[AdoredTV] Pascal vs Maxwell at same clockspeeds, same FLOPS (1080 vs 980 Ti) - Page 28

post #271 of 305
Quote:
Originally Posted by CrazyElf View Post

it does seem like the higher clocks account for most of the improvements.
It is entirely possible that the Samsung/GF 14/20nm process simply isn't good. Certainly the Samsung version of the Apple A8 doesn't give much room for confidence, the TSMC version consumed less power.

A GF version of that process would be even less mature and may have even worse speeds. Compounding the problem, GF has no real experience with discrete GPUs. Ok, they build Carrizo APUs, but that is it.
I wouldn't be so quick to discount TSMC making Vega. There are rumors going around that may be possible.

I actually think that it is plausible. GF has no experience making giant reticle limit sized chips. I'd imagine TSMC would be far better for this. They have years of experience on this.

Outside of OCN, I have heard rumors that they might use TSMC for the big Vega.

There's recent articles out which says VEGA is being made on 14nm finfet.

http://vrworld.com/2016/06/24/amd-celebrates-vega-10-rollout/

Also, the clocks being so low on polaris are mostly because of the architecture itself not the manufacturing node.

GCN itself requires too much volts and generally consumes power too quickly with moderate overclocking. Particularly with the latest versions. Fiji and Tonga displayed the same behavior. Fiji only overclocking to 1450 with ln2 should have given everyone a big flag that finfet was not going to fix these issues. Fiji and Tonga both kicked out double precision performance which should allowed clocks to be driven higher along with being made on a higher performance 28nm node compared to tahiti and hawaii. Nvidia, using the same 28nm process was able to clock up to 2200mhz using ln2. Guess what clocks these are similar to today? Hmmmm. So most of the blame lies with GCN not GF. Also Apples a9 was made using LPE, not LPP. LPP has very similar performance characteristics as 16nm finfet. Although AMD card scale with volts better than Nvidia's chips, after a very small threshold, power consumption explodes. This is why AMD needs water to get that last 10% of overclocking vs what Nvidia could do on air.

One more thing, we better hope it's GCN's and not the process technology of GF thats causing these problems because if it is the process technology, how do you expect Zen to fare vs the best finfet technology in the world at Intel?

One last explanation could be AMD cut R and D on the process design and refinement on all their GPU's as of late because they focused most of their R and D on zen, which is ultimately the smarter thing to do. Zen will make or break AMD, GPU's no so much. The GPU market is too small and is shrinking and doesn't provide enough cash flow to run a company that makes CPU's and GPUs.

These results show not surprise anyone. Pascal and maxwell are a very similar architecture. Atleast IPC didn't go down on a teraflop basis like many suspected. Maxwell already had a strong IPC per teraflop so it's not a surpise Nvidia leveraged this, along with high potential clocks.
Edited by tajoh111 - 7/29/16 at 12:50pm
post #272 of 305
Quote:
Originally Posted by Blameless View Post

The idea that HBM allows less memory to do more is a myth.

HBM is faster, thats it. It doesn't change what VRAM is used for nor does it make up for a lack of VRAM. If anything, all other aspects being equal, HBM(2) would exacerbate the difference between large and small pools of local memory, because the next tier of the memory hierarchy (main system memory, accessed over PCI-E and then the CPU) is that much slower, relatively speaking.

AMD has almost certainly done everything in their power to manage how memory is used on Fiji, but that would apply equally well to non-HBM parts.

Allowing less memory to do more is a generalized statement, that is not what I should be saying. I am, more specifically, saying that HBM allows memory to be used more efficiently, as a "wide" and "slow" technology. Speed is also gained from an interposer as opposed to DDR traces. There is most definitely an inefficient use of the memory buffer:

Quoting CTO Macri at the time:

"If you actually look at frame buffers and how efficient they are and how efficient the drivers are at managing capacities across the resolutions, you'll find that there's a lot that can be done. We do not see 4GB as a limitation that would cause performance bottlenecks. We just need to do a better job managing the capacities. We were getting free capacity, because with [GDDR5] in order to get more bandwidth we needed to make the memory system wider, so the capacities were increasing. As engineers, we always focus on where the bottleneck is. If you're getting capacity, you don't put as much effort into better utilising that capacity. 4GB is more than sufficient. We've had to go do a little bit of investment in order to better utilise the frame buffer, but we're not really seeing a frame buffer capacity [problem]. You'll be blown away by how much [capacity] is wasted."

Moving more data across a wider buffer eleviates the bottleneck, and speed is given through the interposer over trace usage, with power savings from less resistance of current through travel to DDR. This is not the same as "storing more than 4GB textures" but it is is an added benefit of the Fiji's ROPs and HBM's ability to feed it.

Therefore, there is an inherent optimization HBM gives to its neighboring Fiji by enabling a less serialized (and larger) buffer with access to 1GB modules instead of 256/512. So, no there is not a magical 4GB + free storage of larger textures, but the larger buffer gives a better usage VRAM. SoM provides examples of this..

Edit: Let me also say, yes, that is a myth - but not completely. It is also true that GDDR5 today consists of 1GB modules, as well like first rev HBM.

Edit thanks for the clarification.


---


In addition, actually I'm quite sure Fiji was designed from the ground up around HBM. There are no Fiji non-HBM Parts, to my knowledge.
Edited by junkman - 7/29/16 at 3:39pm
post #273 of 305
Quote:
Originally Posted by tajoh111 View Post

Warning: Spoiler! (Click to show)
There's recent articles out which says VEGA is being made on 14nm finfet.

http://vrworld.com/2016/06/24/amd-celebrates-vega-10-rollout/

Also, the clocks being so low on polaris are mostly because of the architecture itself not the manufacturing node.

GCN itself requires too much volts and generally consumes power too quickly with moderate overclocking. Particularly with the latest versions. Fiji and Tonga displayed the same behavior. Fiji only overclocking to 1450 with ln2 should have given everyone a big flag that finfet was not going to fix these issues. Fiji and Tonga both kicked out double precision performance which should allowed clocks to be driven higher along with being made on a higher performance 28nm node compared to tahiti and hawaii. Nvidia, using the same 28nm process was able to clock up to 2200mhz using ln2. Guess what clocks these are similar to today? Hmmmm. So most of the blame lies with GCN not GF. Also Apples a9 was made using LPE, not LPP. LPP has very similar performance characteristics as 16nm finfet. Although AMD card scale with volts better than Nvidia's chips, after a very small threshold, power consumption explodes. This is why AMD needs water to get that last 10% of overclocking vs what Nvidia could do on air.

One more thing, we better hope it's GCN's and not the process technology of GF thats causing these problems because if it is the process technology, how do you expect Zen to fare vs the best finfet technology in the world at Intel? Warning: Spoiler! (Click to show)

One last explanation could be AMD cut R and D on the process design and refinement on all their GPU's as of late because they focused most of their R and D on zen, which is ultimately the smarter thing to do. Zen will make or break AMD, GPU's no so much. The GPU market is too small and is shrinking and doesn't provide enough cash flow to run a company that makes CPU's and GPUs.

These results show not surprise anyone. Pascal and maxwell are a very similar architecture. Atleast IPC didn't go down on a teraflop basis like many suspected. Maxwell already had a strong IPC per teraflop so it's not a surpise Nvidia leveraged this, along with high potential clocks.


Actually, Hawaii has full DP capabilities, just disabled in the firmware. The AMD FirePro W9100 is built on the Hawaii die and has 1/2 DP capability at 2.62 TFLOPs (5.25 TFLOPs of SP performance). The passive S9150 is similar with 1/2 DP. There is also the W8100, a 290 Hawaii professional. DP is just disabled for consumer use on consumer GPUs.


Not as sure about Fury X - they only have released the S9300 X2, which has weak DP performance.

I don't expect Zen to match Skylake at all. I just hope for parity with Sandy Bridge.

Quote:
Originally Posted by Blameless View Post

Overclock any higher-end AMD GPU enough to increase the power consumption 20-30w and you get almost nothing in the way of extra performance. A 10% OC on Hawaii, for example, costs a solid ~50w, and the same trend will apply to any GPU with a similar power target running similarly near the peak of it's headroom.

No doubt that saving power wherever possible is good, but no amount of power savings outside of the GPU proper is going to allow a GPU to scale substantially higher without improvements to the process/yields.
The idea that HBM allows less memory to do more is a myth.

HBM is faster, thats it. It doesn't change what VRAM is used for nor does it make up for a lack of VRAM. If anything, all other aspects being equal, HBM(2) would exacerbate the difference between large and small pools of local memory, because the next tier of the memory hierarchy (main system memory, accessed over PCI-E and then the CPU) is that much slower, relatively speaking.

AMD has almost certainly done everything in their power to manage how memory is used on Fiji, but that would apply equally well to non-HBM parts.


The 4Gb wasn't as big an issue as I initially feared, but for high resolution textures, 4 Gb does not appear adequate. I'd assume that Elite Dangerous, a game you often refer to, would have the same issues with the Fury X as the 290X.

The only thing to do now is to wait for HBM2. I guess maybe AMD could have tried 8 stacks of HBM1, but that would have created other design complexities and they were running out of die space already. It would not have been possible on the Fury X. If they did have the die area, I'd rather they increase the triangle output and the RBEs anyways.
Edited by CrazyElf - 7/30/16 at 11:01am
Trooper Typhoon
(20 items)
 
  
CPUMotherboardGraphicsGraphics
5960X X99A Godlike MSI 1080 Ti Lightning MSI 1080 Ti Lightning 
RAMHard DriveHard DriveHard Drive
G.Skill Trident Z 32 Gb Samsung 850 Pro Samsung SM843T 960 GB Western Digital Caviar Black 2Tb 
Hard DriveOptical DriveCoolingCooling
Samsung SV843 960 GB LG WH14NS40 Cryorig R1 Ultimate 9x Gentle Typhoon 1850 rpm on case 
OSMonitorKeyboardPower
Windows 7 Pro x64 LG 27UD68 Ducky Legend with Vortex PBT Doubleshot Backlit... EVGA 1300W G2 
CaseMouseAudioOther
Cooler Master Storm Trooper Logitech G502 Proteus Asus Xonar Essence STX Lamptron Fanatic Fan Controller  
  hide details  
Reply
Trooper Typhoon
(20 items)
 
  
CPUMotherboardGraphicsGraphics
5960X X99A Godlike MSI 1080 Ti Lightning MSI 1080 Ti Lightning 
RAMHard DriveHard DriveHard Drive
G.Skill Trident Z 32 Gb Samsung 850 Pro Samsung SM843T 960 GB Western Digital Caviar Black 2Tb 
Hard DriveOptical DriveCoolingCooling
Samsung SV843 960 GB LG WH14NS40 Cryorig R1 Ultimate 9x Gentle Typhoon 1850 rpm on case 
OSMonitorKeyboardPower
Windows 7 Pro x64 LG 27UD68 Ducky Legend with Vortex PBT Doubleshot Backlit... EVGA 1300W G2 
CaseMouseAudioOther
Cooler Master Storm Trooper Logitech G502 Proteus Asus Xonar Essence STX Lamptron Fanatic Fan Controller  
  hide details  
Reply
post #274 of 305
Im not sure what language the video is in. Anyone have English version?

And then, well, obviously this -->
Quote:
Originally Posted by CrazyElf View Post


The only thing to do now is to wait for HBM2. I guess maybe AMD could have tried 8 stacks of HBM1, but that would have created other design complexities and they were running out of die space already. It would not have been possible on the Fury X. If they did have the die area, I'd rather they increase the triangle output and the RBEs anyways.

Edited by DMatthewStewart - 7/30/16 at 11:26am
post #275 of 305
Quote:
Originally Posted by CrazyElf View Post

Actually, Hawaii has full DP capabilities, just disabled in the firmware. The AMD FirePro W9100 is built on the Hawaii die and has 1/2 DP capability at 2.62 TFLOPs (5.25 TFLOPs of SP performance). The passive S9150 is similar with 1/2 DP. There is also the W8100, a 290 Hawaii professional. DP is just disabled for consumer use on consumer GPUs.


Not as sure about Fury X - they only have released the S9300 X2, which has weak DP performance.

I don't expect Zen to match Skylake at all. I just hope for parity with Sandy Bridge.
The 4Gb wasn't as big an issue as I initially feared, but for high resolution textures, 4 Gb does not appear adequate. I'd assume that Elite Dangerous, a game you often refer to, would have the same issues with the Fury X as the 290X.

The only thing to do now is to wait for HBM2. I guess maybe AMD could have tried 8 stacks of HBM1, but that would have created other design complexities and they were running out of die space already. It would not have been possible on the Fury X. If they did have the die area, I'd rather they increase the triangle output and the RBEs anyways.

I'll go you one better, the 4GB on Fiji has turned out to be mostly an entire non-issue, even at 4K (as I wondered might be the case before release when everybody was dumping on the 4GB HBM allocation). I have yet to see any Fiji owners that have flat out run short on memory to the point where they actually drop down to less than 10 FPS or so like you would see when it actually runs out. I guess if you're talking about a game with tons of mods and such its possible but I personally have not seen it.

I argued with Blameless for weeks about this and all he has ever said was that HBM had nothing to do with the fact that Fiji hasn't hit any VRAM limitations despite its buffer size but the user experiences would seem to speak for themselves. I have seen claims from other 4GB card owners of memory limits reached (Hawaii for instance) at high resolutions yet Fiji somehow seems to handle it fine. It would seem HBM is doing SOMETHING here (of course this is all anecdotal and if there is evidence to the contrary I'd be interested in seeing it)...
post #276 of 305
Quote:
Originally Posted by Majin SSJ Eric View Post

I'll go you one better, the 4GB on Fiji has turned out to be mostly an entire non-issue, even at 4K (as I wondered might be the case before release when everybody was dumping on the 4GB HBM allocation). I have yet to see any Fiji owners that have flat out run short on memory to the point where they actually drop down to less than 10 FPS or so like you would see when it actually runs out. I guess if you're talking about a game with tons of mods and such its possible but I personally have not seen it.

I argued with Blameless for weeks about this and all he has ever said was that HBM had nothing to do with the fact that Fiji hasn't hit any VRAM limitations despite its buffer size but the user experiences would seem to speak for themselves. I have seen claims from other 4GB card owners of memory limits reached (Hawaii for instance) at high resolutions yet Fiji somehow seems to handle it fine. It would seem HBM is doing SOMETHING here (of course this is all anecdotal and if there is evidence to the contrary I'd be interested in seeing it)...

Pretty sure HBM is nothing to do with it. It just runs at higher bandwidth but doesn't have any specific technology itself to relieve that hard memory limit. AMD put a lot of effort themselves to make sure that Fiji's 4GB limit doesn't become a problem for duration of its shelf life.

http://www.anandtech.com/show/9390/the-amd-radeon-r9-fury-x-review/7
My home PC
(16 items)
 
  
CPUMotherboardGraphicsRAM
AMD Threadripper 1950x Gigabyte Aorus X399 Gaming 7  MSI Geforce GTX 1080ti Gaming X G.Skill DDR4 3600 CL16 
Hard DriveHard DriveCoolingOS
Samsung Evo 840 500GB Samsung 960 Pro 500GB Noctua NH-U14S TR4 Windows 10 Pro 
MonitorMonitorKeyboardPower
Dell U2711 Samsung 55" 4k Corsair K70  EVGA SuperNova G2 1300W 
CaseMouseAudio
Corsair Carbide Air 540 Logitech G502 Denon AVR-X3300W 
  hide details  
Reply
My home PC
(16 items)
 
  
CPUMotherboardGraphicsRAM
AMD Threadripper 1950x Gigabyte Aorus X399 Gaming 7  MSI Geforce GTX 1080ti Gaming X G.Skill DDR4 3600 CL16 
Hard DriveHard DriveCoolingOS
Samsung Evo 840 500GB Samsung 960 Pro 500GB Noctua NH-U14S TR4 Windows 10 Pro 
MonitorMonitorKeyboardPower
Dell U2711 Samsung 55" 4k Corsair K70  EVGA SuperNova G2 1300W 
CaseMouseAudio
Corsair Carbide Air 540 Logitech G502 Denon AVR-X3300W 
  hide details  
Reply
post #277 of 305
Quote:
Originally Posted by Kpjoslee View Post

Pretty sure HBM is nothing to do with it. It just runs at higher bandwidth but doesn't have any specific technology itself to relieve that hard memory limit. AMD put a lot of effort themselves to make sure that Fiji's 4GB limit doesn't become a problem for duration of its shelf life.

http://www.anandtech.com/show/9390/the-amd-radeon-r9-fury-x-review/7

I understand that and certainly could be a correlation/causation deal but the fact remains that other 4GB AMD cards seem to hit memory limits while Fiji cards don't...
post #278 of 305
Quote:
Originally Posted by Majin SSJ Eric View Post

I understand that and certainly could be a correlation/causation deal but the fact remains that other 4GB AMD cards seem to hit memory limits while Fiji cards don't...

GCN 1.1 (aka generation 2, Hawaii) vs GCN 1.2 (aka generation 3, Fiji). Delta color compression would be a significant reason why. It was primarily introduced in hardware with GCN 1.2. Also explains why the RX 480 isn't really bandwidth starved at all either.

Doesn't really have a single thing to do with HBM.
post #279 of 305
Quote:
Originally Posted by junkman View Post

Allowing less memory to do more is a generalized statement, that is not what I should be saying. I am, more specifically, saying that HBM allows memory to be used more efficiently, as a "wide" and "slow" technology. Speed is also gained from an interposer as opposed to DDR traces. There is most definitely an inefficient use of the memory buffer:

Quoting CTO Macri at the time:

"If you actually look at frame buffers and how efficient they are and how efficient the drivers are at managing capacities across the resolutions, you'll find that there's a lot that can be done. We do not see 4GB as a limitation that would cause performance bottlenecks. We just need to do a better job managing the capacities. We were getting free capacity, because with [GDDR5] in order to get more bandwidth we needed to make the memory system wider, so the capacities were increasing. As engineers, we always focus on where the bottleneck is. If you're getting capacity, you don't put as much effort into better utilising that capacity. 4GB is more than sufficient. We've had to go do a little bit of investment in order to better utilise the frame buffer, but we're not really seeing a frame buffer capacity [problem]. You'll be blown away by how much [capacity] is wasted."

Moving more data across a wider buffer eleviates the bottleneck, and speed is given through the interposer over trace usage, with power savings from less resistance of current through travel to DDR. This is not the same as "storing more than 4GB textures" but it is is an added benefit of the Fiji's ROPs and HBM's ability to feed it.

Therefore, there is an inherent optimization HBM gives to its neighboring Fiji by enabling a less serialized (and larger) buffer with access to 1GB modules instead of 256/512. So, no there is not a magical 4GB + free storage of larger textures, but the larger buffer gives a better usage VRAM. SoM provides examples of this..

Edit: Let me also say, yes, that is a myth - but not completely. It is also true that GDDR5 today consists of 1GB modules, as well like first rev HBM.

Edit thanks for the clarification.


---


In addition, actually I'm quite sure Fiji was designed from the ground up around HBM. There are no Fiji non-HBM Parts, to my knowledge.

As one of the few on this forum using fury x crossfire at 4k for almost a year this mirrors my experiences exactly i have YET to see 4gb hbm as a limitation in a game without unreasonable settings. The vram usage is always lower than what gddr5 would use at the same settings.

It still boggles my mind that no ones done in depth testing on this.

I will say it clearly, 4gb hbm (or hbm in general), does not have the same limitations as 4gb gddr5 and cannot be compared directly
Edited by Tgrove - 7/30/16 at 9:47pm
Yokai Onmyoji
(18 items)
 
  
CPUMotherboardGraphicsRAM
I7-4930k @ 4.5GHz ASUS P9X79 Deluxe  Sapphire RX Vega 64 Liquid  G. Skill Ripjaws Z 16GB (4x4) 2400 Mhz  
Hard DriveHard DriveOptical DriveCooling
Samsung 840 Pro 256GB (OS) Samsung 840 Pro 512GB (Games) Lite-on Corsair H100 
OSMonitorKeyboardPower
Windows 7 Ultimate Wasabi Mango UHD490 IPS 49" Freesync 33-60hz Corsair K90 Corsair Ax1200i 
CaseMouseMouse PadAudio
Corsair 500r Razer Naga Dx Racer Arm wrest  Asus Xonar DX 
AudioOther
Altec Lansing 5.1 Razer Nostromo 
  hide details  
Reply
Yokai Onmyoji
(18 items)
 
  
CPUMotherboardGraphicsRAM
I7-4930k @ 4.5GHz ASUS P9X79 Deluxe  Sapphire RX Vega 64 Liquid  G. Skill Ripjaws Z 16GB (4x4) 2400 Mhz  
Hard DriveHard DriveOptical DriveCooling
Samsung 840 Pro 256GB (OS) Samsung 840 Pro 512GB (Games) Lite-on Corsair H100 
OSMonitorKeyboardPower
Windows 7 Ultimate Wasabi Mango UHD490 IPS 49" Freesync 33-60hz Corsair K90 Corsair Ax1200i 
CaseMouseMouse PadAudio
Corsair 500r Razer Naga Dx Racer Arm wrest  Asus Xonar DX 
AudioOther
Altec Lansing 5.1 Razer Nostromo 
  hide details  
Reply
post #280 of 305
Quote:
Originally Posted by Kinaesthetic View Post

GCN 1.1 (aka generation 2, Hawaii) vs GCN 1.2 (aka generation 3, Fiji). Delta color compression would be a significant reason why. It was primarily introduced in hardware with GCN 1.2. Also explains why the RX 480 isn't really bandwidth starved at all either.

Doesn't really have a single thing to do with HBM.

AFAIK most AMD gpu haven't been bandwidth starved for years. They've been choked by weak front end, geometry processing, ROP and cache amounts. Fiji's biggest issue was weak front end, ROPs , and limited cache, hopefully Vega resolves those issues. GCN shaders themselves have the horsepower but AMD pairs them with a front end that can't keep them well fed even with boat loads of bandwidth then squeezes everything out through a thin 64rop straw.

GCN hasn't needed much change on the shader end, it needs a front end that can keep up which it still doesn't have
Edited by Echoa - 7/30/16 at 10:04pm
Toxic-DT
(19 items)
 
Toxic-LT
(12 items)
 
Toxic-SV
(10 items)
 
CPUMotherboardGraphicsRAM
i7 4770k Gigabyte Z87X-UD4H Gigabyte GTX 1060 Gskill Ares 
Hard DriveHard DriveHard DriveOptical Drive
Mushkin Eco3 Kingdian  Corsair LE  Lite On 
CoolingOSMonitorKeyboard
Cryorig H5 Windows 10 pro Dell S2415h Azio MK Retro 
PowerCaseMouseAudio
Seasonic G Series 650w Define R5  Logitech G500s Asus Xonar DSX 
AudioAudioAudio
Monoprice Retro Heaphones Monoprice Vocal Dynamic Mic Edifier r980t  
CPUMotherboardGraphicsRAM
I7 920xm Asus g73jh Radeon HD 6990m 8gb Hynix DDR3 1333 
Hard DriveHard DriveCoolingOS
Seagate Laptop Thin 500gb Corsair Force LE  Stock G73jh Windows 10 
MonitorKeyboardPowerMouse
1600x900 Glossy G73jh Screen G73jh Backlit Keyboard Delta 150w PSU Verbatim Laptop mouse 
CPUMotherboardGraphicsRAM
Xeon e5649 HP z400 FirePro v4800 12gb Gkill Ripjaw 
Hard DriveHard DriveCoolingOS
Hitachi Ultrastar Western Digital 320gb HP High Performance Cooler Ubuntu Gnome 16.10 
PowerCase
HP z400 HP z400 
  hide details  
Reply
Toxic-DT
(19 items)
 
Toxic-LT
(12 items)
 
Toxic-SV
(10 items)
 
CPUMotherboardGraphicsRAM
i7 4770k Gigabyte Z87X-UD4H Gigabyte GTX 1060 Gskill Ares 
Hard DriveHard DriveHard DriveOptical Drive
Mushkin Eco3 Kingdian  Corsair LE  Lite On 
CoolingOSMonitorKeyboard
Cryorig H5 Windows 10 pro Dell S2415h Azio MK Retro 
PowerCaseMouseAudio
Seasonic G Series 650w Define R5  Logitech G500s Asus Xonar DSX 
AudioAudioAudio
Monoprice Retro Heaphones Monoprice Vocal Dynamic Mic Edifier r980t  
CPUMotherboardGraphicsRAM
I7 920xm Asus g73jh Radeon HD 6990m 8gb Hynix DDR3 1333 
Hard DriveHard DriveCoolingOS
Seagate Laptop Thin 500gb Corsair Force LE  Stock G73jh Windows 10 
MonitorKeyboardPowerMouse
1600x900 Glossy G73jh Screen G73jh Backlit Keyboard Delta 150w PSU Verbatim Laptop mouse 
CPUMotherboardGraphicsRAM
Xeon e5649 HP z400 FirePro v4800 12gb Gkill Ripjaw 
Hard DriveHard DriveCoolingOS
Hitachi Ultrastar Western Digital 320gb HP High Performance Cooler Ubuntu Gnome 16.10 
PowerCase
HP z400 HP z400 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
Overclock.net › Forums › Industry News › Hardware News › [AdoredTV] Pascal vs Maxwell at same clockspeeds, same FLOPS (1080 vs 980 Ti)