Topic Title: Low FPS in Tomb Raider with 7970
Topic Summary:
Created On: 03/17/2013 12:55 PM
Status: Post and Reply
Linear : Threading : Single : Branch
1 2 Next Last unread
Search Topic Search Topic
Topic Tools Topic Tools
View similar topics View similar topics
View topic in raw text format. Print this topic.
 03/17/2013 12:55 PM
User is offline View Users Profile Print this message

Author Icon
cliveneo
Grinding Levels

Posts: 121
Joined: 03/09/2008

Hey fellas,

I have a Sapphire 7970 Dual X/W boost. @1100/1500. I have been having issues with tomb raider since day 1. All are fixed now bc of the patches and new 13.3 beta drivers except one critical issue. On three maps Mountain Village, Shantytown and certain area of the Beach my fps drop consideraby. I the peak by the crashed plane in Mountain Village the fps go as low as 25. Especially when u have to save Grim and they scavengers r zip lining down towards u. That was the only combat area that was unplayable. But areas like the plane crash site looking up towards the mountains where the view distance is at its maximum etc. Same in Shantytown. when u r right by the first camp site and look off into the distance 26 FPS. All other areas are perfect 60 fps or occsional dip into the 40s. I posted on two different site with little or no response. Most of the response was from Geforce users. We all know they area having issues. Anyway this game was coded for 7xxx gpus N i have yet to hear a 79xx gamer playing this game stating they were dipping to 25. in fact heard the opposite. My specs are in the sig



-------------------------

FX 8320 @ 4.5_H100i_Sapphire HD7970 1100/1500_ 8gb G.Skill Ripjaw DDR3 1600_Asus Sabertooth 990FX_X-fi Fatl1ty Pro_120GB Corsair Force_64 GB Samsung SSD_1TB WD Black_150GB WD VelociRaptor_150GB WD Raptor_CM HAFX_XFX Pro 850 Watt PSU_Bluray Drive_Samsung 32in HDTV_Sidwinder X6_G500_Z5300 speakers Rig 2: PII X4 925@3.2_Asus M2N86A PLUS_ Gigabyte 560ti 970/1800_4GB DDR2 800 CorsairCMS2_150GB WD Raptor_Ultra 550 Watt PSU_acrylic tech station_VIZO 42in HDTV_logitech g15_G9x



Edited: 03/17/2013 at 01:12 PM by cliveneo
 03/17/2013 01:01 PM
User is offline View Users Profile Print this message

Author Icon
cliveneo
Grinding Levels

Posts: 121
Joined: 03/09/2008

Things ive tried, lowered my oc. reverted the gpu to stock. all the same results the only fix is to lower the overall texture setting to high or lower. Disabling Tressfx, tesselation, FXAA all show no increase in performance which makes me thing there is a bottleneck somewhere. I ran intel burn on maximum for 10 passes, memtest86 for 12 hours, both at stock n OC. ran OCCT on the gpu at stock for 8 hours. Monitored my PSU 12vlt rail nvr dropped below 12.00. nothing is faulty. I dont get why my FPS drop in these areas so much with no fix other than dropping the overall texture setting to either high or medium. I also ran 3dmark 11 and the gpu scores right where it should be. CPU bottleneck maybe? I am not getting a FX 8XXX CPU n a AM3+ mobo for a extra 5-7 fps on average.



-------------------------

FX 8320 @ 4.5_H100i_Sapphire HD7970 1100/1500_ 8gb G.Skill Ripjaw DDR3 1600_Asus Sabertooth 990FX_X-fi Fatl1ty Pro_120GB Corsair Force_64 GB Samsung SSD_1TB WD Black_150GB WD VelociRaptor_150GB WD Raptor_CM HAFX_XFX Pro 850 Watt PSU_Bluray Drive_Samsung 32in HDTV_Sidwinder X6_G500_Z5300 speakers Rig 2: PII X4 925@3.2_Asus M2N86A PLUS_ Gigabyte 560ti 970/1800_4GB DDR2 800 CorsairCMS2_150GB WD Raptor_Ultra 550 Watt PSU_acrylic tech station_VIZO 42in HDTV_logitech g15_G9x



Edited: 03/17/2013 at 01:21 PM by cliveneo
 03/18/2013 08:20 AM
User is offline View Users Profile Print this message

Author Icon
Thanny
Elite

Posts: 1206
Joined: 07/13/2009

The short answer is, you're CPU bound.   I have a 3930K at 4.6GHz, and I am as well at the same kind of areas you're talking about.  I have a pair of 7970's at 1200/1500, and play at 2560x1600.

I ran some tests of a particular view in Mountain Village.  The first thing I did was change the processor affinity mask to see how many real work threads there are.  Performance did not change until I went from three cores to two.  So the game does real work on only three threads, which puts more of a premium on single-core speed.  Getting a FX 8000 series chip, then, would be a step down from what you have now.

Next, using the same view, I set all graphics settings to the lowest levels, and enabled them individually to see which had the biggest impact.  With everything on low, I was CPU bound to 117fps.  No setting had an impact individually save Level of Detail. 

Focusing on Level of Detail, I set everything else to the highest levels, save AA which was set to FXAA, and TressFX which I disabled for more emphasis on CPU load.  Here's what I got with the various Level of Detail settings:

  • Low - 84fps
  • Medium - 82fps
  • Normal - 78fps
  • High - 60fps
  • Ultra - 45fps

GPU usage showed that Normal and up were CPU-limited.  Given that Sandy Bridge is about 40% faster than Phenom II clock for clock, my CPU is about 57% faster than yours, which means your frame rates would look approximately like this:

  • Normal - 50fps
  • High - 38fps
  • Ultra - 29fps

Those are CPU-limited frame rates, and I'm assuming a single 7970 will keep up at 1920x1080.

 



Edited: 03/18/2013 at 08:29 AM by Thanny
 03/18/2013 01:40 PM
User is offline View Users Profile Print this message

Author Icon
zipsi
Case Modder

Posts: 922
Joined: 11/22/2010

I get 45+ fps at all times on ultra (with FXAA) with TressFX on with an i7 3770k oc'd to 4.5ghz.



-------------------------
Intel I7 3770K @ 4.5ghz, Cooler Master Hyper 212 Evo, Gigabyte Z77X-UD3H, Gigabyte HD7970 ghz edition, 4x4gb Kingston HyperX Beast 2133mhz, Seasonic Platinum 860, OCZ Vertex 4 128gb, WD Black 1TB, WD Green 3TB+1.5TB, ASUS Xonar Essence ST, ASUS VE278Q, Windows 3.11
 03/18/2013 05:29 PM
User is offline View Users Profile Print this message

Author Icon
cliveneo
Grinding Levels

Posts: 121
Joined: 03/09/2008

well i dont think its the cpu. i just read a article from techspot. they benched the game at various different settings with geforce cards, amd cpus and intel cpus. they stated and i quote "the game doesnt seem to care what cpu u have as long as it is a modern one with at least for cores." they also stated that with a 7970 mind you there test sys averaged 53 fps. and that increased by 47% when tressfx was disabled. I see NO increase when i disable tressfx. NONE....!!!!!!! at the points i mentioned the only time i see a increase is when i lower the level of detail from ultra to high. thats it. tressfx enabled or disabled 26 to 29fps. yet my gpu benched were it should all other games from flawless. i dont understand. I am seriously just thinking my gpu is junked n sapphire lost my business bc there is no way i should be getting these frame rates



-------------------------

FX 8320 @ 4.5_H100i_Sapphire HD7970 1100/1500_ 8gb G.Skill Ripjaw DDR3 1600_Asus Sabertooth 990FX_X-fi Fatl1ty Pro_120GB Corsair Force_64 GB Samsung SSD_1TB WD Black_150GB WD VelociRaptor_150GB WD Raptor_CM HAFX_XFX Pro 850 Watt PSU_Bluray Drive_Samsung 32in HDTV_Sidwinder X6_G500_Z5300 speakers Rig 2: PII X4 925@3.2_Asus M2N86A PLUS_ Gigabyte 560ti 970/1800_4GB DDR2 800 CorsairCMS2_150GB WD Raptor_Ultra 550 Watt PSU_acrylic tech station_VIZO 42in HDTV_logitech g15_G9x

 03/18/2013 07:33 PM
User is offline View Users Profile Print this message

Author Icon
cliveneo
Grinding Levels

Posts: 121
Joined: 03/09/2008

this is a copy from that post

"Those wanting to play on ultra at 1920x1200 or so will want at least an HD 7870 or its equivalent, which should be a GTX 660 if you change DOF from ultra to high. Although Tomb Raider can cripple our best gear when it's cranked up, it's worth stressing that the game still looks great on the high quality preset and you can expect a stable 60fps with relatively inexpensive cards like the HD 7850 or the GTX 660.

While the game shreds GPUs on higher settings, it doesn't seem to care what processor you have as long as it's a modern chip with at least four cores. We recorded a meager 1fps increase after overclocking the Core i7-3770K from 2.5GHz to 4.5GHz, while the FX-8350 gained a whopping 3fps over the same frequency span. Though it was slower than parts like the i7-3770K, even the Athlon II X2 265 gave playable results."



-------------------------

FX 8320 @ 4.5_H100i_Sapphire HD7970 1100/1500_ 8gb G.Skill Ripjaw DDR3 1600_Asus Sabertooth 990FX_X-fi Fatl1ty Pro_120GB Corsair Force_64 GB Samsung SSD_1TB WD Black_150GB WD VelociRaptor_150GB WD Raptor_CM HAFX_XFX Pro 850 Watt PSU_Bluray Drive_Samsung 32in HDTV_Sidwinder X6_G500_Z5300 speakers Rig 2: PII X4 925@3.2_Asus M2N86A PLUS_ Gigabyte 560ti 970/1800_4GB DDR2 800 CorsairCMS2_150GB WD Raptor_Ultra 550 Watt PSU_acrylic tech station_VIZO 42in HDTV_logitech g15_G9x

 03/18/2013 08:01 PM
User is offline View Users Profile Print this message

Author Icon
cliveneo
Grinding Levels

Posts: 121
Joined: 03/09/2008

here is the link to that post look at the min and max settings. my system should be zippin through this game. there is clearly and obviously something wrong with something in my system. I am going to have to modify my monitoring and see if its my psu dipping or something loosing power somewhere. idk here is link to the site. http://www.techspot.com/review/645-tomb-raider-performance/



-------------------------

FX 8320 @ 4.5_H100i_Sapphire HD7970 1100/1500_ 8gb G.Skill Ripjaw DDR3 1600_Asus Sabertooth 990FX_X-fi Fatl1ty Pro_120GB Corsair Force_64 GB Samsung SSD_1TB WD Black_150GB WD VelociRaptor_150GB WD Raptor_CM HAFX_XFX Pro 850 Watt PSU_Bluray Drive_Samsung 32in HDTV_Sidwinder X6_G500_Z5300 speakers Rig 2: PII X4 925@3.2_Asus M2N86A PLUS_ Gigabyte 560ti 970/1800_4GB DDR2 800 CorsairCMS2_150GB WD Raptor_Ultra 550 Watt PSU_acrylic tech station_VIZO 42in HDTV_logitech g15_G9x

 03/19/2013 06:04 AM
User is offline View Users Profile Print this message

Author Icon
Thanny
Elite

Posts: 1206
Joined: 07/13/2009

Here's another quote from that article you cite:

"We're using Fraps to measure frame rates during 90 seconds of gameplay footage from Tomb Raider’s first level"

In other words, they didn't benchmark any of the large outdoor areas that show the CPU-bound conditions we're talking about.  The article therefore draws a completely incorrect conclusion about the impact of CPU speed.

I'm not guessing here.  I did a specific test and described how anyone can replicate it - find a spot where the FPS drops, then fiddle with the graphics settings while in the game to see how they affect the frame rate, all while monitoring CPU and GPU usage.  Repeat the test yourself.  The spot I picked was just by the Village Overlook camp.  Very similar results are available directly at the Helicopter Hill camp in Shanty Town.

The very fact that you gained nothing by disabling TressFX, a feature that uses the GPU to calculate hair physics, is ample evidence all by itself that you're hitting a CPU bottleneck.  A bottleneck that everyone who plays the game will hit, to a degree that varies by CPU speed.

 

 03/19/2013 02:37 PM
User is offline View Users Profile Print this message

Author Icon
cliveneo
Grinding Levels

Posts: 121
Joined: 03/09/2008

yea i think u r right thanny. after playing with varied cpu speeds it would appear the open areas that i have mention gained a low fps by 3 or 4 bringing it closer to 30. so if i had a fx 8530 and oced it around 4.5 give or take that would increase my min fps to prolly around 35 to 40. I benched every gou demanding game i own. crysis2 max payne 3 etc. n they run flawless so i thing my hardware is secure.

however what i did notice to support ur theory is when the open areas trigger low fps. the gpu load drops from 99% to 60% which would lead me to believe to during that time the game is looking for the cpu to pick up the slack. also yes the 90 sec of the forst lvl make sense bc during the first stage my rig ran at above 55 fps. so i agree with u completely. I need a new cpu. nvr thought a 1090t would be bottle necked, at least not anytime soon. thx thanny for all ur input



-------------------------

FX 8320 @ 4.5_H100i_Sapphire HD7970 1100/1500_ 8gb G.Skill Ripjaw DDR3 1600_Asus Sabertooth 990FX_X-fi Fatl1ty Pro_120GB Corsair Force_64 GB Samsung SSD_1TB WD Black_150GB WD VelociRaptor_150GB WD Raptor_CM HAFX_XFX Pro 850 Watt PSU_Bluray Drive_Samsung 32in HDTV_Sidwinder X6_G500_Z5300 speakers Rig 2: PII X4 925@3.2_Asus M2N86A PLUS_ Gigabyte 560ti 970/1800_4GB DDR2 800 CorsairCMS2_150GB WD Raptor_Ultra 550 Watt PSU_acrylic tech station_VIZO 42in HDTV_logitech g15_G9x

 03/19/2013 03:35 PM
User is offline View Users Profile Print this message

Author Icon
GreatExpectations
Peon

Posts: 11
Joined: 03/15/2013

@Thanny: If you say its CPU bound how can one explain that single core usage barely reaches 80% instead of over 90%? I did some tests like underclocking my Phenom be 955 to 2100MHZ (The game still didnt max out a core) and i barely lost any frames (from 20 to 16 on the overlook of mountain camp). Overclocking by 400 gained me 4 fps. I also gain no frames unless i lower Level of Detail. I can leave everything else on Ultra specs. Can it be that the graphics card doesn't get enough input? Is that a hyper-transfer bottleneck?  

I'm running a Phenom II BE 955 3.2ghz (now at 3.6) and a HD 7870 XT.

Further more: I tried some more games like Far Cry 3 and Tomb Raider seems to be the only game where I have this CPU bottleneck. The only reasonable upgrade for me in this case would be to go i7 or is there an AMD cpu i could reasonably upgrade to?



Edited: 03/19/2013 at 03:41 PM by GreatExpectations
 03/19/2013 07:16 PM
User is offline View Users Profile Print this message

Author Icon
Mime
Forum Moderator

Posts: 576
Joined: 08/28/2012

Single core usage?... I'm confused... The game is multithreaded with more than one worker thread, so you're only seeing part of the story by looking at a single thread.

Also... dropping from 20 to 16 is a different story than dropping from 60 to 56.  It's a reduction of 20% at a point where the system is already under heavy load.  If it wasn't, then the frame rate would be higher than 20. 



-------------------------

Do not meddle in the affairs of archers, for they are subtle and quick to anger.
Post Count: +8510
Troll Hunter

The opinions expressed above do not represent those of Advanced Micro Devices or any of their affiliates.

 03/20/2013 01:10 AM
User is offline View Users Profile Print this message

Author Icon
GreatExpectations
Peon

Posts: 11
Joined: 03/15/2013

Yeah I thought about that after I wrote it. I just heard to identify cpu bottlenecks that are not due to badly optimized software you'll check if 1 or 2 of the cores reach close to 100%. I monitored all 4 of my cores naturally. 1 of them reached around 80 one 60 and the other two stayed around 30. I'm just interested in hearing what that means, if like the northbridge is too slow if it could be something else (like RAM - which I don't think).

 03/20/2013 02:21 AM
User is offline View Users Profile Print this message

Author Icon
Mime
Forum Moderator

Posts: 576
Joined: 08/28/2012

It means the game is CPU limited, in some areas. 

Why your machine doesn't behave exactly the same as someone else's machine is a different question.  You've got different hardware, maybe different operating systems, different software running in the background, and any other number of variables which can skew the comparison.



-------------------------

Do not meddle in the affairs of archers, for they are subtle and quick to anger.
Post Count: +8510
Troll Hunter

The opinions expressed above do not represent those of Advanced Micro Devices or any of their affiliates.

 03/20/2013 11:57 AM
User is offline View Users Profile Print this message

Author Icon
cliveneo
Grinding Levels

Posts: 121
Joined: 03/09/2008

yes GE we are in a slight disadvantage in the CPU front. the only upgrade i can make is to a 83xx FX series and im sure my minimum fps will increase to the upper 30s to 40. but for one game... nah ill wait until AMDs next CPU line. everything else i game dont hit a bottleneck so plus i completed TR with 100% so moving on



-------------------------

FX 8320 @ 4.5_H100i_Sapphire HD7970 1100/1500_ 8gb G.Skill Ripjaw DDR3 1600_Asus Sabertooth 990FX_X-fi Fatl1ty Pro_120GB Corsair Force_64 GB Samsung SSD_1TB WD Black_150GB WD VelociRaptor_150GB WD Raptor_CM HAFX_XFX Pro 850 Watt PSU_Bluray Drive_Samsung 32in HDTV_Sidwinder X6_G500_Z5300 speakers Rig 2: PII X4 925@3.2_Asus M2N86A PLUS_ Gigabyte 560ti 970/1800_4GB DDR2 800 CorsairCMS2_150GB WD Raptor_Ultra 550 Watt PSU_acrylic tech station_VIZO 42in HDTV_logitech g15_G9x

 03/20/2013 04:27 PM
User is offline View Users Profile Print this message

Author Icon
Mime
Forum Moderator

Posts: 576
Joined: 08/28/2012

Another thing to consider is that the Bulldozer based chips have weaker per-thread performance, so you would have to overclock the chip in order to make up for that.  If you happen to get stuck with a dud that won't OC... then you're kinda screwed.



-------------------------

Do not meddle in the affairs of archers, for they are subtle and quick to anger.
Post Count: +8510
Troll Hunter

The opinions expressed above do not represent those of Advanced Micro Devices or any of their affiliates.

 03/20/2013 06:56 PM
User is offline View Users Profile Print this message

Author Icon
neo5555
Case Modder

Posts: 905
Joined: 08/13/2008

Originally posted by: cliveneo   nah ill wait until AMDs next CPU line.

 

Unfortunately, the future of performance CPU's from AMD is very grim..  They've pretty much given up on competing with Intel on the performance front and instead they are focusing on budget APU's..   The FX-8350 is apparently the last of AMD's performance chips, so if you want something faster you will need to join the dark side my friend...

 

 

Originally posted by: Mime If you happen to get stuck with a dud that won't OC... then you're kinda screwed.

 

 

 

Thats very unlikely to be case. Most FX chips overclock really well and the only limiting factor is a motherboard that can handle it properly.



-------------------------

Pics of my rig
i7 4770k, custom water cooling, 8Gig Corsair Dom 2133 9-11-10-27, ASrock Z87 OC Formula, SilverStone ST-1500 PSU, Creative X-Fi Platinum, Logitech Z-5500D , 2xHIS 7970 CF, Asus VW246H Monitor, Win 7 64Bit HP, Corsair 800D Case

 03/20/2013 07:06 PM
User is offline View Users Profile Print this message

Author Icon
Mime
Forum Moderator

Posts: 576
Joined: 08/28/2012

Duds are rare... yeah, but it happens.  If you're going to invest more in the machine, trying to OC a 1090T a little more seems like a better idea than buying a new FX and needing to OC that just to get back to where the 1090T was in the first place.



-------------------------

Do not meddle in the affairs of archers, for they are subtle and quick to anger.
Post Count: +8510
Troll Hunter

The opinions expressed above do not represent those of Advanced Micro Devices or any of their affiliates.

 03/20/2013 07:56 PM
User is offline View Users Profile Print this message

Author Icon
Thanny
Elite

Posts: 1206
Joined: 07/13/2009

Originally posted by: GreatExpectations @Thanny: If you say its CPU bound how can one explain that single core usage barely reaches 80% instead of over 90%? I did some tests like underclocking my Phenom be 955 to 2100MHZ (The game still didnt max out a core) and i barely lost any frames (from 20 to 16 on the overlook of mountain camp). Overclocking by 400 gained me 4 fps. I also gain no frames unless i lower Level of Detail. I can leave everything else on Ultra specs. Can it be that the graphics card doesn't get enough input? Is that a hyper-transfer bottleneck? 

 

This is going to be a long answer.

Measured CPU usage involves picking a time interval, then reporting the percentage of the CPU cycles over that interval that were spent executing instructions versus spent idle.  So say you have a single processing core, and measure usage over a one-second interval.  During that interval, you run a single-threaded task that takes as much CPU time as it can get, but lasts only half a second.  The reported CPU usage would be 50%, even though it was actually 100% for the duration of that thread's execution.

Take the same test and run it on a system with two processing cores.  The OS will normally schedule the thread to run alternately on each core in a more or less round robin ***** (uh, since when is "f a s h i o n" a word to censor?).  The reported usage would then be 25% for each core.  Again, even though the task was using every CPU cycle for each slice of time that it ran on each processing core.

Now consider a single thread that's running constantly and wants as much CPU time as it can get, with actual usage monitoring that uses a much smaller time interval for sampling.  On a single-core system, usage will be reported as 100%.  On a dual-core system, the per-core usage will fluctuate pretty wildly, but always add up to 100% across the two cores, and probably average around 50% each.

Closer to the topic at hand, now assume you have three work threads (as is the case with Tomb Raider) running at maximum speed.  Let's call that 300% CPU usage, since a single thread at full usage on a single processing core would be 100%.  You have four processor cores, which means a capacity of 400% processor usage.  Put that 300% load on your 400% system and you will see the usage of each core hovering around 75%.  You will see the same usage figure after underclocking your system.

More generally again, this can change if the program chooses to override the default scheduling and sets an affinity mask for either the whole process or individual threads.  If I write a program which uses two threads for doing work and then calls SetProcessAffinityMask() to exclude cores 1 and 2 of a four-core CPU, the reported usage for cores 3 and 4 will accurately reflect the load my program is imposing.

One more complication is Hyperthreading.  With two logical processors for each physical core, the reported usage is basically half the actual usage.  So if core 1 is at 50% load, according to something like Core Temp, that really means it's at 100% load, but one of the integer units is idle.  Each logical core gets its own graph in Task Manager, but the CPU usage figure is scaled by total logical core count.  So a single thread running at full tilt on a four-core processor with Hyperthreading would be reported as using 12% of the CPU (truncated from the actual value of 12.5%).  On your Phenom II, the figure is 25%.

Finally, Level of Detail is definitely what hits the CPU hardest.  While that may sound like something that should impact the GPU, what it really means is that the number of objects and models in the world go up, which in turns requires the CPU to do more calculations to build and place those models.  The polygon count almost certainly increases as well, which would mean a higher load on the GPU if all else were equal.  But it's not, because the CPU can't keep up, so the GPU load will actually go down as fewer frames are set up by the CPU for rendering on the GPU.

 

Further more: I tried some more games like Far Cry 3 and Tomb Raider seems to be the only game where I have this CPU bottleneck. The only reasonable upgrade for me in this case would be to go i7 or is there an AMD cpu i could reasonably upgrade to?

 

 

Tomb Raider is certainly not the only game that hits a CPU-shaped brick wall.  An even better example is Assassin's Creed III, where several areas are CPU-limited to the low 30's on my 4.6GHz 3930K.  Among older games, Crysis is very CPU-heavy, but the ratio of CPU load to GPU load is much better.  Most games do have a better fit between CPU and GPU requirements.

As far as upgrade possibilities go, for poorly threaded games that put gigantic loads on the CPU, the only viable path for improvement is to an Intel i7.  Even then, as you can see from the numbers I've reported, there are still badly CPU-bound scenarios to contend with. 

Moving to Bulldozer (FX 8150) would be a huge step down, as Phenom II is much faster clock for clock.  Piledriver (FX 8350) is better than Bulldozer, but still slower than Phenom II, so you'd need a clock speed advantage just to break even.



Edited: 03/20/2013 at 08:20 PM by Thanny
 03/20/2013 08:14 PM
User is offline View Users Profile Print this message

Author Icon
Thanny
Elite

Posts: 1206
Joined: 07/13/2009

Originally posted by: cliveneo yea i think u r right thanny. after playing with varied cpu speeds it would appear the open areas that i have mention gained a low fps by 3 or 4 bringing it closer to 30. so if i had a fx 8530 and oced it around 4.5 give or take that would increase my min fps to prolly around 35 to 40.

 

 

Piledriver is roughly 5% slower than Phenom II on a clock for clock basis.  A FX 8350 at 4.5GHz would be less than 5% faster than your 1090T at 4.1GHz.  So where you're getting 30fps now, you'd be getting about 31fps with a FX 8350. 

Your only path to real improvement is to jump back to Intel, where pretty much any i7 at 4.5GHz would get you from 30fps to about 45fps.  Lowering the Level of Detail setting is a much cheaper fix.

 

 03/20/2013 08:34 PM
User is offline View Users Profile Print this message

Author Icon
Mime
Forum Moderator

Posts: 576
Joined: 08/28/2012

Yeah I started to write a post explaining scheduling and how there can still be a CPU bottleneck without reported usage reaching 100%, but clearly I was too lazy to do that.  Instead I just said "it happens because it happens". 

This kind of scheduling exists in order to give the illusion of being able to run as many applications as we want simultaneously.  Even though a typical desktop CPU has only a few physical cores there's often a few hundred threads in some stage of execution at any point in time.  The scheduler of an OS is what makes that work and appear(mostly) seamless.  They're devious little buggers sometimes, but useful.



-------------------------

Do not meddle in the affairs of archers, for they are subtle and quick to anger.
Post Count: +8510
Troll Hunter

The opinions expressed above do not represent those of Advanced Micro Devices or any of their affiliates.

Statistics
80951 users are registered to the AMD Support and Game forum.
There are currently 1 users logged in.

FuseTalk Hosting Executive Plan v3.2 - © 1999-2014 FuseTalk Inc. All rights reserved.