Overclock.net › Forums › Industry News › Hardware News › [Various] Futuremark Releases 3DMark Time Spy DirectX 12 Benchmark
New Posts  All Forums:Forum Nav:

[Various] Futuremark Releases 3DMark Time Spy DirectX 12 Benchmark - Page 72

post #711 of 772
Just noticed something curious in my last bench attempt. I recently reenabled C&Q and C1E on my FX-8320e after tuning tuning up my voltages and Ram timings with it locked at 4.0GHz. I have never had throttling with the System locked at 4.0GHz with those power saving features disabled, and in limited stress testing and benchmarking since i enabled them, i have also never had seen throttling.

That is until today. I noticed in the 1st GFX test under Time Spy my cores throttled down to low power state on multiple occasions. I ran several other tests including Firestrike and saw none, nor was there any on the 2nd test or the CPU test on TS. Just seemed curious to me. I may try to replicate it a few times and post a screen shot if anyone wants to see what i'm talking about.
post #712 of 772
Disable APM
FX
(7 items)
 
  
CPUMotherboardGraphicsRAM
FX-8320@4.4Ghz M5A99FX PRO R2.0 AMD Radeon R9 290X Patriot Memory  
Hard DriveCoolingMonitor
Samsung 840 Raijintek Ereboss iiyama X4071UHSU (4K) 
  hide details  
Reply
FX
(7 items)
 
  
CPUMotherboardGraphicsRAM
FX-8320@4.4Ghz M5A99FX PRO R2.0 AMD Radeon R9 290X Patriot Memory  
Hard DriveCoolingMonitor
Samsung 840 Raijintek Ereboss iiyama X4071UHSU (4K) 
  hide details  
Reply
post #713 of 772
Quote:
Originally Posted by gapottberg View Post

Just noticed something curious in my last bench attempt. I recently reenabled C&Q and C1E on my FX-8320e after tuning tuning up my voltages and Ram timings with it locked at 4.0GHz. I have never had throttling with the System locked at 4.0GHz with those power saving features disabled, and in limited stress testing and benchmarking since i enabled them, i have also never had seen throttling.

That is until today. I noticed in the 1st GFX test under Time Spy my cores throttled down to low power state on multiple occasions. I ran several other tests including Firestrike and saw none, nor was there any on the 2nd test or the CPU test on TS. Just seemed curious to me. I may try to replicate it a few times and post a screen shot if anyone wants to see what i'm talking about.

Time Spy graphics tests are not CPU tests. They are specifically designed so that the CPU would not be a bottleneck.

It can still be if you pair something silly like GTX1080 on a Core 2 Duo, but on anything reasonably modern, the CPU indeed can go down to lower power states during the Graphics Tests.

CPU test then jumps on the CPU and beats it down until it is crying (while GPU(s) basically idle).

By design, so we can isolate the performance of each.

Fire Strike also had a Combined Test that do both, but we omitted it from Time Spy, partially due to the fact that Time Spy development cycle was already long enough, and partially because in DX12 the CPU is no longer such a big deal - you will be hard pressed to find games that end up CPU limited due to the much lower driver overhead. So making a fair and useful Combined Test was a non-trivial task to say the least. We may give it another go in a future 3DMark test, but it was left out here.
post #714 of 772
Please help me guys. Why my physicial score is so bad, from where comes that bad score only 3700..
http://www.3dmark.com/3dm/13472460
post #715 of 772
Are there still some unanswered technical questions about TimeSpy? You can contact Futuremark, email is in the article they published as a response.

They do answer, although it's rather a politically correct answers without giving much useful information. Summary is they did invest a lot of time and feedback from GPU makers who had seen the code many times and their main goal is not to push each GPU to it's max but rather to emulate and estimate future performance of GPUs in upcoming games by estimating the amount of effort game developers may give when making and optimizing their engines.
As such none of Futuremark's benchmarks can be used to compare maximum possible hardware performance of different GPUs but rather be used as a wanna be game benchmark. IMHO the development is still affected by the domination of Nvidia on the market, especially when it comes to game developers which then give feedback to Futuremark about what they do/don't do/invest time into when making their engines and as such Futuremark emulates the current poor state of game development which is mostly optimized for Nvidia cards.


If there is some never ending question here and you can't get answer from Futuremark, notify me, I'll try asking them.
post #716 of 772
Quote:
Originally Posted by KeepWalkinG View Post

Please help me guys. Why my physicial score is so bad, from where comes that bad score only 3700..
http://www.3dmark.com/3dm/13472460

I may be wrong but it says you ran it on an INTEL Graphics card tongue.gif!
 
QUEEN OF BLADES
(15 items)
 
ASUS U6Sg
(15 items)
 
CPUMotherboardGraphicsRAM
I7 920 REV D0@4.2 HT Asus Rampage II Extreme CFX: ASUS R9 270X DCUII 6GB DDR3 XMP CL6 TRIDENT 
Hard DriveOptical DriveCoolingOS
Vertex 2 180 GB+RAID 0 2 * 1 TB SAMSUNG Sony Optiarc Labelflash Thermalright Venoumous X W7 X64 Ultimate RETAIL 
MonitorKeyboardPowerCase
Dell U2410 IPS Logitech Corsair CMPSU-850TX Antec 1200 
MouseMouse PadAudio
Logitech HAMA Xonar STX / Auzen Forte 
CPUMotherboardGraphicsRAM
XEON 5650X @4.45 GHz  Rampage III Formula PCGH CFX: ASUS R9 280X DCUII 16GB G-SKILL CL9@ 1643 MHz 
Hard DriveOptical DriveCoolingOS
1x 850 Pro 256 GB + 1xWD 500 GB + 2x3TB WD RED Plextor PX-L89OSA Thermalright VX W10 X64 Ultimate RETAIL 
MonitorKeyboardPowerCase
ACER B203W Razer Deathstalker Corsair CMPSU-750TX Antec 902 
MouseMouse PadAudio
Logitech G800S Logitech Gaming  Creative Titanium HD + FiiO E17 
CPUMotherboardGraphicsRAM
Intel Core 2 Duo : T8300 Asus U6000Sg Series Notebook NVIDIA 9300 GS 4GB Kingston CL4 
Hard DriveOptical DriveCoolingOS
Hitachi Ultrastar Asus POS Stock Asus with MX3 W7 x64 Ultimate RETAIL 
MonitorAudio
Asus 12.1 " LED Creative X-Fi EX54 
  hide details  
Reply
 
QUEEN OF BLADES
(15 items)
 
ASUS U6Sg
(15 items)
 
CPUMotherboardGraphicsRAM
I7 920 REV D0@4.2 HT Asus Rampage II Extreme CFX: ASUS R9 270X DCUII 6GB DDR3 XMP CL6 TRIDENT 
Hard DriveOptical DriveCoolingOS
Vertex 2 180 GB+RAID 0 2 * 1 TB SAMSUNG Sony Optiarc Labelflash Thermalright Venoumous X W7 X64 Ultimate RETAIL 
MonitorKeyboardPowerCase
Dell U2410 IPS Logitech Corsair CMPSU-850TX Antec 1200 
MouseMouse PadAudio
Logitech HAMA Xonar STX / Auzen Forte 
CPUMotherboardGraphicsRAM
XEON 5650X @4.45 GHz  Rampage III Formula PCGH CFX: ASUS R9 280X DCUII 16GB G-SKILL CL9@ 1643 MHz 
Hard DriveOptical DriveCoolingOS
1x 850 Pro 256 GB + 1xWD 500 GB + 2x3TB WD RED Plextor PX-L89OSA Thermalright VX W10 X64 Ultimate RETAIL 
MonitorKeyboardPowerCase
ACER B203W Razer Deathstalker Corsair CMPSU-750TX Antec 902 
MouseMouse PadAudio
Logitech G800S Logitech Gaming  Creative Titanium HD + FiiO E17 
CPUMotherboardGraphicsRAM
Intel Core 2 Duo : T8300 Asus U6000Sg Series Notebook NVIDIA 9300 GS 4GB Kingston CL4 
Hard DriveOptical DriveCoolingOS
Hitachi Ultrastar Asus POS Stock Asus with MX3 W7 x64 Ultimate RETAIL 
MonitorAudio
Asus 12.1 " LED Creative X-Fi EX54 
  hide details  
Reply
post #717 of 772
Quote:
Originally Posted by JackCY View Post

Are there still some unanswered technical questions about TimeSpy? You can contact Futuremark, email is in the article they published as a response.

They do answer, although it's rather a politically correct answers without giving much useful information. Summary is they did invest a lot of time and feedback from GPU makers who had seen the code many times and their main goal is not to push each GPU to it's max but rather to emulate and estimate future performance of GPUs in upcoming games by estimating the amount of effort game developers may give when making and optimizing their engines.
As such none of Futuremark's benchmarks can be used to compare maximum possible hardware performance of different GPUs but rather be used as a wanna be game benchmark. IMHO the development is still affected by the domination of Nvidia on the market, especially when it comes to game developers which then give feedback to Futuremark about what they do/don't do/invest time into when making their engines and as such Futuremark emulates the current poor state of game development which is mostly optimized for Nvidia cards.


If there is some never ending question here and you can't get answer from Futuremark, notify me, I'll try asking them.

All I got was, we (Futuremark) have been doing benchmarks for a long time, so we know how the future games will be developed... That doesn't make any sense to me. Although I doubt that Futurmark will be able to provide further granularity, it would be interesting to know how many and which game developers actually provided input to Futuremark, and what set of assumptions were used in framing the inquiry.
My guess is that Futuremark looked at a handful of current DX 12 games that are out now but were in the pipeline when Time spy was still in development, and Futuremark just used that as the basis of this benchmark, and then proceeded to make a sweeping bold claim about "next 1-3 years of game development"...And, this is the part that I find most ridiculous.. Lol

Edit: Oh, and by the way, that "interview" was a lay up, with the interviewer often leading the witness..lol seemed more of a PR piece than anything else...
Edited by provost - 7/21/16 at 5:33am
Simplicity
(11 items)
 
Apotheosis
(10 items)
 
 
CPUMotherboardGraphicsRAM
4770k Asus Z87 Pro TBD Corsair Vengeance (2x8GB) DDR3 1600 RAM 
OSMonitorKeyboardPower
Windows 7 Pro Dell U2713HM Alienware TactX gaming Seasonic 850W Gold  
CaseMouse
Cooler Master HAF XB Alienware TactX premium mouse 
  hide details  
Reply
Simplicity
(11 items)
 
Apotheosis
(10 items)
 
 
CPUMotherboardGraphicsRAM
4770k Asus Z87 Pro TBD Corsair Vengeance (2x8GB) DDR3 1600 RAM 
OSMonitorKeyboardPower
Windows 7 Pro Dell U2713HM Alienware TactX gaming Seasonic 850W Gold  
CaseMouse
Cooler Master HAF XB Alienware TactX premium mouse 
  hide details  
Reply
post #718 of 772
My poor 3D Club Series 13 7790 pushed to its limits 1345/1800 core +80mV * bios from Asus 7790 for 1,5V on Vrams wink.gif * higher OC or more mV will end with throttle I believe TDP limit because VRM heatsink is not hot wink.gif
initial run http://www.3dmark.com/3dm/13474483 but MS windows start do some **** in background and occupied my CPU frown.gif
next runs cannot achieve this same GPU score ;/ http://www.3dmark.com/3dm/13475689

To handle more heat I add some heastinks on heatpipe and put new thermal compound, and I tightened the screws so much until I heard the screech of springs biggrin.gif
Temp decrease from 85C *100% rpm to 75C *72% rpm in Furmark wink.gif

Edited by Rabit - 7/21/16 at 4:50am
Cheap VR bulid :)
(16 items)
 
   
CPUMotherboardGraphicsRAM
i5 3470 @ 4Ghz P67A-GD65 (B3) Gtx 970 Windforce Hynix HMT351U6BFR8C-H9 
Hard DriveHard DriveHard DriveCooling
ADATA SP550 Samsung HD502HJ 500GB ocz vertex 2 Zalman CNPS 10X Quiet 
OSMonitorKeyboardPower
Windows 10 AOC 2236Swa Techno-X Silentium Deus M1 550W 
CaseMouseMouse PadOther
Aerocool GT Advance Case Maxtek mouse China cheap :P Sharkoon Shark Zone H40 
CPUMotherboardGraphicsRAM
X4 860k ASUS A88XM-A Club 3d Series 13 7790 Corsair CMZ8GX3M2A1600C8R 
Hard DriveHard DriveOptical DriveCooling
Samsung HD502HJ 500GB ADATA SP550 LG GH22LS50 Scythe Mugen 
OSMonitorPower
Windows 7 Home 64 LG Flatron E2242 Silentium Deus M1 550W 
  hide details  
Reply
Cheap VR bulid :)
(16 items)
 
   
CPUMotherboardGraphicsRAM
i5 3470 @ 4Ghz P67A-GD65 (B3) Gtx 970 Windforce Hynix HMT351U6BFR8C-H9 
Hard DriveHard DriveHard DriveCooling
ADATA SP550 Samsung HD502HJ 500GB ocz vertex 2 Zalman CNPS 10X Quiet 
OSMonitorKeyboardPower
Windows 10 AOC 2236Swa Techno-X Silentium Deus M1 550W 
CaseMouseMouse PadOther
Aerocool GT Advance Case Maxtek mouse China cheap :P Sharkoon Shark Zone H40 
CPUMotherboardGraphicsRAM
X4 860k ASUS A88XM-A Club 3d Series 13 7790 Corsair CMZ8GX3M2A1600C8R 
Hard DriveHard DriveOptical DriveCooling
Samsung HD502HJ 500GB ADATA SP550 LG GH22LS50 Scythe Mugen 
OSMonitorPower
Windows 7 Home 64 LG Flatron E2242 Silentium Deus M1 550W 
  hide details  
Reply
post #719 of 772
Quote:
Originally Posted by ZealotKi11er View Post

Yeah but 1070 seem to be 15-20% faster. Hey I love to see why 1070 is faster. I am just worried about myself. I consider Nvidia as much as AMD and do not want to end up with a 980 Ti like card which would cost me $1000 CAD + and end up at the bottom with no explanation.

When I looked at the price of the 980Ti while shopping, a lot of them were available from as low as $599 CAD. Even the Fury X is at $690 at NCIX.

I ended up getting a GTX 1070 because .. you know, new tech ! tongue.gif
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
i7-6700K @ 4.7GHz 1.395v ASUS Z170-A Gigabyte GTX 1070 G1 Gaming G.SKILL Ripjaws V Series DDR4 3000MHZ 16GB 
Hard DriveCoolingOSMonitor
Samsung EVO 850 250GB Noctua NH-D15S Windows 10 Pro BenQ XL2720Z 
KeyboardPowerCaseMouse
Corsair K70 EVGA SuperNOVA 850w G3  Phanteks Evolv ATX Corsair Scimitar PRO RGB / Logitech G502  
  hide details  
Reply
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
i7-6700K @ 4.7GHz 1.395v ASUS Z170-A Gigabyte GTX 1070 G1 Gaming G.SKILL Ripjaws V Series DDR4 3000MHZ 16GB 
Hard DriveCoolingOSMonitor
Samsung EVO 850 250GB Noctua NH-D15S Windows 10 Pro BenQ XL2720Z 
KeyboardPowerCaseMouse
Corsair K70 EVGA SuperNOVA 850w G3  Phanteks Evolv ATX Corsair Scimitar PRO RGB / Logitech G502  
  hide details  
Reply
post #720 of 772
Quote:
Originally Posted by provost View Post

All I got was, we (Futuremark) have been doing benchmarks for a long time, so we know how the future games will be developed... That doesn't make any sense to me. Although I doubt that Futurmark will be able to provide further granularity, it would be interesting to know how many and which game developers actually provided input to Futuremark, and what set of assumptions were used in framing the inquiry.
My guess is that Futuremark looked at a handful of current DX 12 games that are out now but were in the pipeline when Time spy was still in development, and Futuremark just used that as the basis of this benchmark, and then proceeded to make a sweeping bold claim about "next 1-3 years of game development"...And, this is the part that I find most ridiculous.. Lol

Edit: Oh, and by the way, that "interview" was a lay up, with the interviewer often leading the witness..lol seemed more of a PR piece than anything else...
You mean the PcPer interview on YT yesterday?

Yes they've said TS was in development for a few years since DX12 didn't even exist and as such being released along with DX12 games that are barely optimized it does make one question who did they ask for input when it comes to game devs, it seems to me it's been mostly DX11/OGL devs that only started to use LL APIs later. All in all they do an estimate not a scientific comparison.

I'll see if I get a reply from them again and any new technical question comes up. Overall they are explaining how DX12 works and what choices they made and why and who had the chance to review those choices in code.
They've used the one shoe fits all approach which means using a sort of approach that runs on all HW, the thing is Nvidia HW is optimized to run fast when using this approach and while AMD HW is more flexible and can be used in different ways to bring out more performance those options are not available and trash performance on other HW so it is not used. This doesn't even have to be in terms of async and feature levels at all, you can see the same differences in DX11/OGL when optimized for the path that works for all/Nvidia or one that takes advantages of Nvidia especially (looking at you Project Cars) or one that takes advantages of AMD hardware which tends to run poor on Nvidia HW. The same happens in DX12 and Vulkan regardless of what features or async level is used. You feed the GPU improperly which causes it to trash the caches or have unnecessary idle time and any HW is gonna run poor and it don't matter what API is used.

IMHO Nvidia dominates the PC game developers with their support and market share which means many games are optimized toward Nvidia HW which runs worse on AMD and doesn't use any HW advantages AMD HW has. This is often seen in benchmarks where Nvidia optimized titles are CPU speed dependent when using any GPU, which is in contrast to AMD optimized titles that offer almost equal performance no matter how fast CPU is used.
I bet it has to do with Nvidia preferring single threaded GPU workload upload where as AMD approach happily runs multi threaded.

As such Nvidia enjoys limited HW capabilities using it's lock on the market to keep it that way, it allows them to create more power efficient HW as it is more specialized and less flexible. Until Nvidia changes their architecture to be able to run things in HW in parallel I doubt AMD will get out of this crappy market situation anytime soon on PC unless there will be a huge amount of ports from Consoles or game devs being forced by parent companies to release games on both PC and consoles in such a case it makes more sense to develop on AMD HW and optimize it for that.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
Overclock.net › Forums › Industry News › Hardware News › [Various] Futuremark Releases 3DMark Time Spy DirectX 12 Benchmark