Overclock.net › Forums › AMD › AMD CPUs › [Failed Promises or Fine Wine?] Discussion on the gains from software optimizations for AMD based Systems over time
New Posts  All Forums:Forum Nav:

[Failed Promises or Fine Wine?] Discussion on the gains from software optimizations for AMD based Systems over time - Page 2

post #11 of 16
Quote:
Originally Posted by OzzyRuleZ View Post

I had a 8350 @ 4.7ghz before switching to a Ryzen 1700X. It did game well. I had zero issues with it and was happy with it's performance. Especially with Mantle/Vulkan/DX12 titles. It still lives on in a friends kid's PC and he is super happy with it. However one big thing I noticed with Ryzen vs FX is heat. FX generated a very large amount of heat compared to Ryzen. It could comfortably warm my room in the winter time, in the summer it was unbearable. The air exhausted from my PC was hot. Now with Ryzen it is only a few degrees above ambient. So I am getting better performance and with a considerable amount less heat.

Man I remember I had to turn the heater off whenever I played games like bf4 on my 4.6ghz 8320.
post #12 of 16
I don't think that Bulldozer is a good comparison. Bulldozer was so slow in IPC and so bad in power efficiency that it was not worth optimizing for. By contrast, Zen is pretty fast in IPC and power efficient.

We have already seen optimizations with Ryzen on 2 games - Total War: Warhammer and Ashes of Singularity. In other words, Ashes has already received the optimization that you claimed it would not get.



Unless you want to pay a lot more money for that 3 or so extra fps (maybe 4-5 if the 6900K is also at DDR4 3200), then Ryzen is your CPU to get. Oh, and that assumes that there won't be any further improvements. That's not a good assumption because Oxide says they are in beta still, plus there might be more room. So really you aren't getting much in workstation or gaming for your money with X99 now. I guess the 6900K getting 4.2 to 4.4 GHz is something compared to 3.9 to 4.1 GHz with Ryzen, but even then that's not much.

http://www.overclock.net/t/1627243/hardware-fr-maj-total-war-warhammer-pour-ryzen-total-war-warhammer-just-got-a-ryzen-patch/0_100#post_25985575
http://www.overclock.net/t/1626820/pcper-ashes-of-the-singularity-gets-ryzen-performance-update/0_100














More on Ashes of SIngularity

See this too:


https://www.extremetech.com/gaming/246840-new-ashes-singularity-update-substantially-boosts-ryzen-performance

Everyone agrees it is a bit slower on Ryzen right now than Broadwell E. Before the optimization, (see image above), at 1080p, there was a huge penalty that disappeared almost by 4k.



PC Gamer too, after optimization sees huge gains.



Quote:
“Every processor is different on how you tune it, and Ryzen gave us some new data points on optimization,” Oxide’s Dan Baker told PCWorld. “We’ve invested thousands of hours tuning Intel CPUs to get every last bit of performance out of them, but comparatively little time so far on Ryzen.”

Baker said Oxide wanted to get the beta out to the world so users could at least see the potential. Oxide’s CEO also said (in a statement released by AMD), “as good as AMD Ryzen is right now—and it’s remarkably fast—we’ve already seen that we can tweak games like Ashes of the Singularity to take even more advantage of its impressive core count and processing power. AMD Ryzen brings resources to the table that will change what people will come to expect from a PC gaming experience.”

Oxide isn’t the only one starting to tune for Ryzen. Bethesda also said it had formed a partnership with AMD to optimize and support the company’s CPUs and GPUs.

http://pcworld.com/article/3185466/hardware/heres-proof-that-ryzen-can-benefit-from-optimized-game-code.html


In other words, they have more work to do (hence my comment above that it was a bad assumption that the gap won't be closed entirely). Once the full optimizations are out, we will be GPU limited, not CPU limited. All we need to know is who else is going to optimize on Ryzen. Combined with the faster RAM, we should be in the clear.

Furthermore, Stardock/Oxide, Creative Assembly, and now Bethesda have already committed to making games that are Ryzen optimized. I would not be surprised if EA/DICE or other vendors decided to optimize on Ryzen.















Other Thoughts


I think it is fair to assume that other games will get it. Oxide games has shared their optimization data for Ashes already too. It's possible we may see further gains with Ryzen soon.

With faster RAM, most of the gaming gap goes away as well. Other optimizations such as turning off the HPET, disabling SMT, etc, can also yield gains on some games. I have written previously about this.
http://www.overclock.net/t/1625187/the-ryzen-gaming-performance-gap-is-mostly-gone/0_100

I do not believe any games were optimized for Bulldozer. By contrast, we have already seem games being optimized. In any even, the games gap is no longer very big. We are GPU, not CPU bottlenecked in most titles. That's where we want to be.


Zen is pretty much trading blows otherwise with workstation benchmarks, exceeding Broadwell E in many cases. By comparison, Bulldozer is simply not worth optimizing for. It sucked. It's power efficiency was terrible. Zen by contrast has pretty good power efficiency - actually at moderate clocks even better than Intel's. Right now the 32 core monster that AMD is sampling is clocking at 3.1 GHz with 180W TDP - independent tests have shown with underclocking that perhaps 3.2 or 3.3 GHz might be possible with 180W TDP. That's pretty good in my book - in fact it's going to be a big competition for Intel.

There are some issues with Zen that need to be addressed. The queues need to be widened. The Infinity Fabric needs more bandwidth (hope this does not hurt Vega too much). They should decouple the Fabric, RAM, and core speeds (like on Intel's Skylake). Finally there are a few IPC improvements they could give it that could get it to Skylake level. Even better, with Samsung's 14LPU process, they might be able to get the clockspeeds up a few hundred MHz - maybe more with more High Density Library Optimizations (keep in mind Ryzen is the first time AMD launched with HDL on 14nm).
Edited by CrazyElf - 4/8/17 at 9:08pm
Trooper Typhoon
(20 items)
 
  
CPUMotherboardGraphicsGraphics
5960X X99A Godlike MSI r9 290X Lightning  MSI r9 290X Lightning 
RAMHard DriveHard DriveHard Drive
G.Skill Trident Z 32 Gb Samsung SM843T 960 GB Western Digital Caviar Black 2Tb Samsung 850 Pro 
Hard DriveOptical DriveCoolingCooling
Samsung SV843 960 GB LG WH14NS40 Cryorig R1 Ultimate 9x Gentle Typhoon 1850 rpm on case 
OSMonitorKeyboardPower
Windows 7 Pro x64 Korean 27" 2560x1440 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 r9 290X Lightning  MSI r9 290X Lightning 
RAMHard DriveHard DriveHard Drive
G.Skill Trident Z 32 Gb Samsung SM843T 960 GB Western Digital Caviar Black 2Tb Samsung 850 Pro 
Hard DriveOptical DriveCoolingCooling
Samsung SV843 960 GB LG WH14NS40 Cryorig R1 Ultimate 9x Gentle Typhoon 1850 rpm on case 
OSMonitorKeyboardPower
Windows 7 Pro x64 Korean 27" 2560x1440 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 #13 of 16
Thread Starter 
@CrazyElf: Um...i appreciate the enthusiasm, but i'm just not sure who you are replying to. The point of the thread is to share with others your personal real world experience on how your FX platform is aging well. Benchmarks and evidence are appreciated, but nothing you shared shows much related to the topic of the thread. There was never any intention to compare Bulldozer, or Piledriver, or FX in general; to anything else...Ryzen, Intel or otherwise. Merely to show that a good FX system still has relevant gaming chops, and may have a few years of life left in them.

I too am very optimistic about Ryzen's future, but id appreciate it if that were left to the Ryzen threads already going. A link to one of them might be more appropriate in the future if you really want to share it in this thread.

The one thing i disagree with is you opinion on optimization being pointless on FX. The reality is that optimization never needed to be FX specific for it to thrive...just multi-threaded specific to remove its perceived lack of compute power. The power was always there, just not coded for. The sad part is that better multi threading improved performance on Intel chips as well. Its not like the gains were only AMD's to reap. It is a shame it has taken so long to get the market to move past a Quad core design we have had since back in what, 2006? Pretty stoked Ryzen will help move us into the future and make 8 threads mainstream, and as it does software will take advantage of it more and more. As that happens, even with slower older less capable cores, FX will continue to age better and better.
post #14 of 16
From personal experience, yes FXs has aged very well i bought mine FX6300 back on December '12 by that time there was many reviewers who would bash FXs (and they still are). I was and am really happy about performance too bad i wasn't into synthetics back then so i don't have any kind of benchmarks the only thing i remember when 3Dmark introduced API overhead i was getting about ~750k drawcalls in ST and ~650k in MT (DX11) and now im about +1,4million in ST (yes i know this thread is focused towards DX12, Vulkan but the gap in DX11 is pretty visible biggrin.gif ).

Games like crysis 3,cs:go,bioshock inf,far cry,gta v,metro ll,dota2 run alot better by CPU perspective also DX12 and Vulkan are big 'updates' for FXs, as a DOOM fan was impressed about the performance my CPU would deliver when using vulkan though i cannot say the same for ROTTR and dx12 but still it run better than dx11, if these apis used properly the future of FXs still bright even if they're like 4? years old are capable of decent gaming. Overclocking these CPUs is the best part i usually use fsb+MP to reach better clocks in lower voltages and this gives me better single threaded performance (at least for me) i have tweaked my ram timings which also helps because over time i've observe FXs are hungry for high bandwidth and low latency (IMC is the saddest story about FXs imo biggrin.gif ).

Lately i was thinking moving to Ryzen 5 because i do a lot of photoshop+video editing but decide to stay a year or so with my lovely 6300 and buy a new 4k monitor with better sRGB-Adobe colors and better PP-low ms biggrin.gif (currently 6300 in my country sells for 80€ if i can find 8300 for lower i would consider it as those 2 extra core/threads would help me a lot with my work).

Here's some tips from me for FX owners: Warning: Spoiler! (Click to show)
1)Find a good cooling solution, wraith cooler or EVO 212 is a good-to go.

2)Overclock it to 4,5GHz or more. Usually you can achieve it with 1,45V (disable all power-saving features before overclock), use Prime95 or IBT AVX, after keeping an eye on temperatures (CPU sites says 62.5c is the max-safe but people on forums arguing they can reach safely 70c (

3)Buy high speed-low CL ram. Try and play with Ram Timings and Ram Speeds, memory controller is the real problem of FXs. Overclocking Bus speed can help with ram but it affects other factors like CPU speed, HT link speed (Really great if you having PCI-e Bandwidth problems). In my opinion getting the best of RAM in these chips is must for me, you can clearly see gaming performance increase.

4)Don't use MSAA depending in game (some affect CPU) instead use DSR-VSR or SMAA if available, also lower Draw/View Distance if available. Using DSR-VSR is the best way (i don't recommend FXAA really hate it, makes the whole screen Blurry its a no-no for me) you can also try some 3rd party like Reshade-Sweetfx to enable SMAA. (SMAA is the future biggrin.gif ).

5)Make sure windows wont park cores. Open Performance monitor to see if parked cores exist. Am not sure if windows 10 park cores but W7-W8.1 tend to do that there's registry key to disable that future you can search on google there also programs for that purpose.

6)Check your Motherboard BIOS ver. So we come to this, as already many know updating your BIOS is rare and somewhat dangerous and most of tech gurus don't recommending unless you are having problems, personally i was having bad downclocks and my ram wouldn't change to XMP profile updating to the newest bios solve all of my problems i could reach 4.8Ghz with 1.45V and tweak my RAM.
Oxezz-PC
(12 items)
 
R5 Laptop
(11 items)
 
 
CPUMotherboardGraphicsRAM
A8-7410 FT3b R5 Graphics  x1 4GB 1333Mhz 
Hard DriveOptical DriveCoolingOS
SSD 500GB YES Pipes lol W10 Home 64Bit 
MonitorKeyboardPower
1920x1080 poor colors Integrated Power Cord 
  hide details  
Reply
Oxezz-PC
(12 items)
 
R5 Laptop
(11 items)
 
 
CPUMotherboardGraphicsRAM
A8-7410 FT3b R5 Graphics  x1 4GB 1333Mhz 
Hard DriveOptical DriveCoolingOS
SSD 500GB YES Pipes lol W10 Home 64Bit 
MonitorKeyboardPower
1920x1080 poor colors Integrated Power Cord 
  hide details  
Reply
post #15 of 16
Quote:
Originally Posted by OzzyRuleZ View Post

I had a 8350 @ 4.7ghz before switching to a Ryzen 1700X. It did game well. I had zero issues with it and was happy with it's performance. Especially with Mantle/Vulkan/DX12 titles. It still lives on in a friends kid's PC and he is super happy with it. However one big thing I noticed with Ryzen vs FX is heat. FX generated a very large amount of heat compared to Ryzen. It could comfortably warm my room in the winter time, in the summer it was unbearable. The air exhausted from my PC was hot. Now with Ryzen it is only a few degrees above ambient. So I am getting better performance and with a considerable amount less heat.

We both started out from the same place with a Gigabyte UD3 rev 3 motherboard. Most of that heat is coming from that UD3, not the FX. The back side cover would get so hot it would burn my fingers. No ideer how hot the VRMs behind it were, but placing my hand in the exhaust stream made me think it was like sticking my hand in an oven. When I changed motherboards the temps dropped dramatically and now when running at 1.536 v it puts out a lot less heat than that UD3 did at 1.4 v.

No one is saying that FX is a world beater. We are saying it is doing much better than it was 4 years ago and will continue to perform well for years to come. FX is aging well. I'm frankly more interested in Vega than Ryzen as this rig is gpu bottlenecked today.
Silver & Black
(17 items)
 
Blackie
(17 items)
 
 
CPUMotherboardGraphicsRAM
AMD Ryzen 1700 Biostar X370 GT7 Sapphire Nitro Fury G. Skill Ripjaws ddr4 3200 16 GB 4x4 
Hard DriveHard DriveOptical DriveCooling
Samsung 850 EVO 500gb SSD Western Digital 500gb  Memorex Lightscribe dvd Thermalright Silver Arrow IB-E Extreme 
OSMonitorKeyboardPower
Win 10 64bit Pixio 27" 1440P yes Seasonic 850 watt  
CaseMouseMouse PadAudio
Silverstone Redline RL05 Evoluent vertical mouse yes Creative X-FI  
Other
Sennheiser headphones 
CPUMotherboardGraphicsRAM
AMD FX8370 @ 5 ghz ASUS Sabertooth Sapphire Fury G Skill F3-14900CL9Q-16GBXL 
Hard DriveHard DriveOptical DriveCooling
Western Digital Samsung 850 EVO Memorex Lightscribe dvd Thermalright Silver Arrow IB-E Extreme 
OSMonitorKeyboardPower
Win X 64 bit Pixio 27" 1440P  yes Seasonic SS-850KM Active PFC F3 
CaseMouseMouse PadAudio
Silverstone Redline RL02b Evoluent VM4 yes Creative Soundblaster Z  
Audio
Sennheiser 428 headphones 
  hide details  
Reply
Silver & Black
(17 items)
 
Blackie
(17 items)
 
 
CPUMotherboardGraphicsRAM
AMD Ryzen 1700 Biostar X370 GT7 Sapphire Nitro Fury G. Skill Ripjaws ddr4 3200 16 GB 4x4 
Hard DriveHard DriveOptical DriveCooling
Samsung 850 EVO 500gb SSD Western Digital 500gb  Memorex Lightscribe dvd Thermalright Silver Arrow IB-E Extreme 
OSMonitorKeyboardPower
Win 10 64bit Pixio 27" 1440P yes Seasonic 850 watt  
CaseMouseMouse PadAudio
Silverstone Redline RL05 Evoluent vertical mouse yes Creative X-FI  
Other
Sennheiser headphones 
CPUMotherboardGraphicsRAM
AMD FX8370 @ 5 ghz ASUS Sabertooth Sapphire Fury G Skill F3-14900CL9Q-16GBXL 
Hard DriveHard DriveOptical DriveCooling
Western Digital Samsung 850 EVO Memorex Lightscribe dvd Thermalright Silver Arrow IB-E Extreme 
OSMonitorKeyboardPower
Win X 64 bit Pixio 27" 1440P  yes Seasonic SS-850KM Active PFC F3 
CaseMouseMouse PadAudio
Silverstone Redline RL02b Evoluent VM4 yes Creative Soundblaster Z  
Audio
Sennheiser 428 headphones 
  hide details  
Reply
post #16 of 16
Thread Starter 
Here is something i have been meaning to post for some time, and something to keep in mind when looking at all those canned benchmark results yous see flying around with Ryzens release. You really need to know the context of what it is you are looking at, and what it means. Below is a screen shot of MSI ABs data tracking during a gaming session in AoS:E, 6 player Large map i recently did. Had literally hundreds and maybe even over a thousand or so units out at a time between all the teams at various points in this game. Needless to say it is representative of a typical game for me, and it played nearly flawlessly at 1080p High on my 60hz monitor.

After a brief Idle session, I then ran the 3 min Canned GPU Benchmark and tracked it as well. Take a look and see what you notice differs in the two runs (separated by yellow marker)...




First off, the GPU usage is fairly similar in both cases, so that's good to see; though it's not a perfect mirror of conditions so i left it up. I did removed graphs like ram and memory usage that turned out to be identical in each run, but what really caught my eye was CPU usage. Obviously the Benchmark is providing the CPU with a scenario that as atypical of actual game play, and much more demanding than what is actually being experienced under typical gaming conditions. Maybe a 14 team large army clash would achieve results more similar to the benchmark, but not in most normal play scenarios would you ever find conditions that demanding (it doesn't same CPU usage as above).

This is just the GPU benchmark, and as we have seen from my own and other graphs from other reviewers...even an FX-83XX with a mild/moderate OC that is being pegged out at 100% usage during the bench, will still giving you above 30fps minimums and close to 50fps averages on a R9-390X. If you absolutely overload it by using the AoS:E CPU test it drops your AVG to about 22fps, but once again it is completely atypical of any actual in game scenario i have seen from my time playing the game.

What is missing when a lot of people throw these benchmarks out as a review is context. Neither the GPU or CPU benchmarks for this game actually simulate anything resembling an in game experience. Anytime a reviewer is using the benches results to directly or indirectly to demonstrate, or even just speculate what actual in game performance is like; they are flat out wrong. A benchmark is only as good as its context. Clearly the context for the AoS:E benchmark is "given a scenario that far exceeds typical gameplay, what is ones performance for particular hardware capable of" ; NOT... "Running this benchmark is a good indication of what your in game experience will be like."

Knowing this context as allowed me to use AoS:Es in game benchmark for a wide variety of things. It has become one of my favorite tools for stress testing my machine and one type of measurement i track for performance gains through OCing and tweaking my system. I never use it however to quickly ascribe how my in game experience will look in AoS or any other game. They are not the same thing. There may very well be games out there with "good" bench marking tools that do just that. And i believe that is what the vast majority of people believe they are and should be doing, but often they do not. However, just because this one does not, does not make it less "good" per say, as long as you understand what it is delivering you and apply it properly like i and many others do.

The final point i want to make in all this babbling is that clearly the in game experience for me has been better than the benchmark would suggest it should be. The evidence suggests that it is because the benchmark is extremely more demanding than what we have in game now, and yet even the more demanding scenario often lends itself to what would appear to be a very playable experience on my hardware.

So yeah, my FX has the chops for this game as it is today (@1080/1440p, 60hz), and also likely still has the chops for more demanding versions of similar games when coded in DX12 or similar APIs that take full advantage of the multi-core/multi-thread design my FX leverages. I don't see this platform getting retired anytime soon for me, and if anything it will become my son's main rig when i do move up for years to come.

[Edit] Fixed my OP. thumb.gif
Edited by gapottberg - 4/11/17 at 2:23pm
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD CPUs
  • [Failed Promises or Fine Wine?] Discussion on the gains from software optimizations for AMD based Systems over time
Overclock.net › Forums › AMD › AMD CPUs › [Failed Promises or Fine Wine?] Discussion on the gains from software optimizations for AMD based Systems over time