Overclock.net banner
261 - 277 of 277 Posts

·
Iconoclast
Joined
·
32,327 Posts
Quote:
Originally Posted by 8051 View Post

Where in the Techspot article does it state that the minimum FPS they observed happened once in a 3 hour period? What if the minimum FPS they recorded happened every time?
It says minimum FPS.

Doesn't matter if it happens every run, both the duration and the frequency are unknowable from their figures. Without knowing more than just the minimum FPS reached, you have no way of knowing what the actual impact on gameplay is. It's reasonable to assume that the results with lower average and much lower minimum FPS are resulting in a worse experience, but the minimum FPS itself means almost nothing and cannot be reasonably weighed more than the average FPS without more information.

All of which is why I want to see 0.1% figures...the average for the worst 0.1% of frames, and how it differs from the worst 1%, is far more more telling than the single worst frame (which is what minimum fps is).

Here is an example of what I'm talking about:

titan-x-1080pultra-frametime-p1.png


Minimum frame rate is about 9, but the worst 0.1% is 31 and worst 1% 73; which says that other that one or or two stutters, the game probably ran perfectly fine. If the worst 0.1% was a lower figure, it would be obvious that stutters were a frequent occurance, but know the all time worst frame rate was 9 doesn't say anything other than how long the longest individual stutter was.
 

·
*cough* Stock *cough*
Joined
·
2,790 Posts
Min fps is like a huge skip. It does matter to many users.

With DDR3 getting quickly replaced, I do not plan to replace my memory(gskill 1866 stuff), but may try to see if it will overclock to see how it feels.
 

·
Iconoclast
Joined
·
32,327 Posts
Quote:
Originally Posted by Nukemaster View Post

Min fps is like a huge skip. It does matter to many users.
It's unduly weighted by even more. Many games will see horrible absolute minimum FPS no matter how fast the system or it's memory, while others will experience major gameplay issues without showing poor minimum FPS.

It's not difficult to get more detailed figures and I find it rather annoying when a test only lists minimum and average.
 

·
Banned
Joined
·
5,087 Posts
Quote:
Originally Posted by Blameless View Post

It says minimum FPS.

Doesn't matter if it happens every run, both the duration and the frequency are unknowable from their figures. Without knowing more than just the minimum FPS reached, you have no way of knowing what the actual impact on gameplay is. It's reasonable to assume that the results with lower average and much lower minimum FPS are resulting in a worse experience, but the minimum FPS itself means almost nothing and cannot be reasonably weighed more than the average FPS without more information.

All of which is why I want to see 0.1% figures...the average for the worst 0.1% of frames, and how it differs from the worst 1%, is far more more telling than the single worst frame (which is what minimum fps is).
Yes they could've provided more in-depth results, but the increase in minimum FPS in their tests would be noticeable by anyone in their test run, because it's not insignificant.
 

·
Iconoclast
Joined
·
32,327 Posts
Quote:
Originally Posted by 8051 View Post

Yes they could've provided more in-depth results, but the increase in minimum FPS in their tests would be noticeable by anyone in their test run, because it's not insignificant.
A significant change in a variable that itself may or may not be significant doesn't say much. This is a fundamental problem with looking at what can easily be a one-time outlier. For example, I've done tests with GTA V where the setup giving a better overall gaming experience had an order of magnitude lower minimum FPS; one setup had a small loading pause at the beginning of a specific area transistion that knocked fps to virtually 0, while the other setup only fell to about 10 fps, but was still worse everywhere else.

In the Techspot review, the minimum FPS figures don't reveal much more than the average fps figures. The difference in average FPS proportionally similar to minimum, except in the case of the i3...a difference that may come down to having less cache, or could just an anomaly (no way to tell for sure without more info). That's almost a best case scenario...it should be easy to see how a single runt frame could give a misleading idea of the impact of any number of variables.
 

·
Registered
Joined
·
627 Posts
Quote:
Originally Posted by Scotty99 View Post

I dont believe for a second memory speed can raise min fps by half, i am tempted to buy a 2400 kit of ddr3 just to test against my 1600 cas8 set. Id actually put a lot of money if someone wanted to bet me, i dont think id barely get any gains.
if you buy a 2400c11 kit the difference will be minimal
1600c8 is a lot faster than 1600c11 or 1333c9
rather than buying a kit maybe try overclocking your current kit there is a good chance that with 1.65v it will do much higher speeds with relaxed timings

some tests i ran here on fallout 4 arma3 r6 siege
smaller min fps variance than techspot in fallout 4 but still significant
http://www.overclock.net/t/1611359/3770k-vs-6700k-in-fallout-4-arma-3-rainbow-six-siege-with-core-cache-memory-scaling-ddr3-1600c11-2133c9-ddr4-2133c15-3000c12-4000c17

this is a run through fallout 4 on the 3770k as you can see the min fps are not just a spike and are very consistent
unfortunately due to the test being short with low to high fps just being a second out starting\stopping the tests though the average results a fair bit


this is a older arma3 test where i looked more closely at timings
min fps in arma3 are rather spiky though so not very consistent
average fps variance was under 1fps between runs
https://forums.bistudio.com/topic/156993-arma-3-cpu-vs-ram-performance-comparison-1600-2133-up-to-15-fps-gain/
 

·
Registered
Joined
·
753 Posts
Guys i need help , just upgraded into Team Vulcan 2400mhz 16GB CL11
But my system simply won't boot with XMP 2400.

Tried 1.65V/1.75V with VCCSA 1.0V and VCCIO 1.150V still won't boot.

Try to manually set 2400mzh and Timings, same no luck.

I can only boot with Speed 2000MHZ CL9/10 1.65V and VCCSA/VCCIO at auto..
 

·
Registered
Joined
·
6,053 Posts
Quote:
Originally Posted by zipper17 View Post

Guys i need help , just upgraded into Team Vulcan 2400mhz 16GB CL11
But my system simply won't boot with XMP 2400.

Tried 1.65V/1.75V with VCCSA 1.0V and VCCIO 1.150V still won't boot.

Try to manually set 2400mzh and Timings, same no luck.

I can only boot with Speed 2000MHZ CL9/10 1.65V and VCCSA/VCCIO at auto..
Is this the sig rig? Very possible that particular Biostar board doesn't have compatibility with this RAM.

Also from personal experience Sandy and Ivy Bridge didn't benefit from faster RAM as much. Where as Haswell really started to show the potential of faster RAM.
 

·
Registered
Joined
·
3,412 Posts
Quote:
Originally Posted by zipper17 View Post

Guys i need help , just upgraded into Team Vulcan 2400mhz 16GB CL11
But my system simply won't boot with XMP 2400.

Tried 1.65V/1.75V with VCCSA 1.0V and VCCIO 1.150V still won't boot.

Try to manually set 2400mzh and Timings, same no luck.

I can only boot with Speed 2000MHZ CL9/10 1.65V and VCCSA/VCCIO at auto..
Loosen the timing a bit and try again. The speeds and timing are not always doable. It depends on the IMC quality in your CPU a lot. Also the mobo plays a small factor as well.
 

·
Registered
Joined
·
753 Posts
Quote:
Originally Posted by iRUSH View Post

Is this the sig rig? Very possible that particular Biostar board doesn't have compatibility with this RAM.

Also from personal experience Sandy and Ivy Bridge didn't benefit from faster RAM as much. Where as Haswell really started to show the potential of faster RAM.
Yes pc spec on my sig,

It could be, but there is 1 review at OCN my mobo TZ77XE4 running 2400-2600mhz just fine however with Corsair Dominator.
http://www.overclock.net/t/1261185/test-for-biostar-tz77xe4-loading-with-core-i7-3770k-overclocking-4-8g
Quote:
Originally Posted by Prophet4NO1 View Post

Loosen the timing a bit and try again. The speeds and timing are not always doable. It depends on the IMC quality in your CPU a lot. Also the mobo plays a small factor as well.
Tried some random timing CL9/CL10-10-10/CL11-11-11/CL12-12-12/CL13, CL 11-13-13-35{same XMp timings}, CL14/15 etc
VCCIO/VTT as high as 1.2V, 1.150V, 1.075V etc
VCCSA/IMC as high as 1.2V, 1.1V, 1.0V
VDIMM 1.65-1.75V {didn't try 1.8V though, afraid for damaging component}
Internal PLL Enabled
3570K AT default or 3570K @4.5GHZ

2400/2133mhz simply won't boot.

it could be bad mobo/defective RAM, incompability, or something else...

However Switch back to my old ram Vengeance 8Gb 1600 CL9,
it can Boot at 2200MHZ CL 9/10 @1.6V Just fine, but didn't test stability, however tried 2400MHZ also the same wont boot.

as for right now I'm running All 4 sticks together Total 24GB with 1600MHZ CL 9-9-9-24 -2T @1.5V, 3570K @4.5GHZ, seems fine. tried small testing with HCIdesign MemTest ~100coverage no error. What do you suggest a software for stability testing?
 

·
Registered
Joined
·
3,412 Posts
Quote:
Originally Posted by zipper17 View Post

Yes pc spec on my sig,

It could be, but there is 1 review at OCN my mobo TZ77XE4 running 2400-2600mhz just fine however with Corsair Dominator.
http://www.overclock.net/t/1261185/test-for-biostar-tz77xe4-loading-with-core-i7-3770k-overclocking-4-8g
Tried some random timing CL9/CL10-10-10/CL11-11-11/CL12-12-12/CL13, CL 11-13-13-35{same XMp timings}, CL14/15 etc
VCCIO/VTT as high as 1.2V, 1.150V, 1.075V etc
VCCSA/IMC as high as 1.2V, 1.1V, 1.0V
VDIMM 1.65-1.75V {didn't try 1.8V though, afraid for damaging component}
Internal PLL Enabled
3570K AT default or 3570K @4.5GHZ

2400/2133mhz simply won't boot.

it could be bad mobo/defective RAM, incompability, or something else...

However Switch back to my old ram Vengeance 8Gb 1600 CL9,
it can Boot at 2200MHZ CL 9/10 @1.6V Just fine, but didn't test stability, however tried 2400MHZ also the same wont boot.

as for right now I'm running All 4 sticks together Total 24GB with 1600MHZ CL 9-9-9-24 -2T @1.5V, 3570K @4.5GHZ, seems fine. tried small testing with HCIdesign MemTest ~100coverage no error. What do you suggest a software for stability testing?
Seems to me like the IMC in the CPOU is just not strong enough for 2400. You may be able to play with voltages to get it running, but it's probably not worth it.
 

·
Registered
Joined
·
753 Posts
Quote:
Originally Posted by Prophet4NO1 View Post

Seems to me like the IMC in the CPOU is just not strong enough for 2400. You may be able to play with voltages to get it running, but it's probably not worth it.
hmm you mean i need to bump cpu vcore for Higher clock RAM, it maybe has some effect?
hmm okay.
 

·
Registered
Joined
·
3,412 Posts
Quote:
Originally Posted by zipper17 View Post

hmm you mean i need to bump cpu vcore for Higher clock RAM, it maybe has some effect?
hmm okay.
No, NOT Vcore. You need to bump the voltages related to the IMC. VTT and VCCSA typically. Those are related to the non core specific functions of the CPU. VTT used to be called system agent. VCCSA is the IMC it's self.
 

·
Registered
Joined
·
753 Posts
Quote:
Originally Posted by Prophet4NO1 View Post

No, NOT Vcore. You need to bump the voltages related to the IMC. VTT and VCCSA typically. Those are related to the non core specific functions of the CPU. VTT used to be called system agent. VCCSA is the IMC it's self.
yeah already tried bumping VCCSA/VCCIO step by step seems no luck.
 

·
Registered
Joined
·
753 Posts
After trial&error for a week, now I can boot up with 2133MHZ CL10-12-12-32 2T @1.65V VCCSA/VCCIO auto, paired with 3570K @4.5GHZ, +0.08V, 50% LLC!

Run Prime95 min/max FFT size 864K for about +90minutes, no error, no whea error, max temp 83C. I hope it would stay stable.
hair.gif


1344K = Vcore
448K = Vrin/Input
512-576 = Cache/Uncore
672-720K = VTT
768K = Agent/IMC
800K = Vdimm/Timings
864K = All

Still wont boot with 2400(@1200mhz), btw here is Aida64 SPD report:


My Mobo Only support TRFC timings limited to 255, when Vulcan 2400(1200mhz) seems needs TRFC at 314, and CommandRate @3T.
Is this might be the culprit? I found several problems too in google people with TRFC is limited at 255 on their mobo.
would someone let me know whats your max TRFC in your mobo? and whats your aida64 SPD report looks like? just curious thanks.
if some timing locked at certain value in mobo while RAM need a higher timing value, would it cause this Boot issue?

Btw Playing Hitman Marrakesh in Intense Market Area, i can feel minimum framerates is increased.
smile.gif

3570K same at 4.5GHZ but;
With 1600MHZ I can see the lowest is 49.7 FPS...
With 2133MHZ I never see 49.7 FPS anymore, Lowest is +51FPS...

with 16GB RAM also possible to open Firefox+Hitman at the same time without insufficient memory occurs like on 8GB RAM, in my experience.
 

·
Iconoclast
Joined
·
32,327 Posts
Virtually, nothing actually needs command rate 3, but tRFC can be a problem. AUTO often selects the correct figure, even on boards where the manual range is not high enough.
 
261 - 277 of 277 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top