chiliwili69 Posted June 30, 2017 Posted June 30, 2017 (edited) I have just acquired a brand new 4790K since it is the top processor for the 1150 socket. Also acquired a basic air cooler (Sycthe Katana 4, since my case is microATX) because I am planning to do some moderate overclocking. It is my first "K" processor and I have never done overclocking before (I didn´t have the need to do it when playing at monitor but in VR I need more). I have spent some hours reading about this "science" or "art". It is a whole world of freaks things (I saw people using liquid nitrogen to cool the CPU!!) that I am really lazy to learn, I just prefer to use my limited time playing BOS. I have learnt that I can run the OC in default auto mode, and the system is running the cores in the 4.2 - 4.4 range. I also have learnt that I can force to manual mode and run all the cores to a fix multiplier rate (44 x 100MHz) and try do put the lowest stable Vcore voltage, so there will be less heat to be removed and the temperatures will be lower. Nobody recommend the auto mode since it uses a higher voltage than needed. I have also updated my BIOS, my Chipset and downloaded a ton of tools: MSI Command Center, XTU, Open Hardware Monitor, CPU-Z, GPU-Z... This weekend I expect to have some free time to start to do some testing with the OC thing. I measure how the BOS performance increases. Is there any particular trick I should be aware about? Any recommended clock speed/Vcore for 4790K with BOS with air cooling? If I run some cores at 4.2 and others at 4.4, is BOS selecting the fastest cores to run in? Edit: I have just found a nice table for speed/Vcore http://www.overclock.net/t/1411077/haswell-overclocking-guide-with-statistics Edited July 1, 2017 by chiliwili69
=TBAS=Sshadow14 Posted June 30, 2017 Posted June 30, 2017 wait did i miss something..IL2 is maxing out your 4790K? NO WAY EVER!!Lol mate on the stock speeds the game will used about 10-20% CPU usage in MULTIPLAYER and maybe 5-15% in Singleplayer.Even with my FX8350 @ 4.4Ghz game has never used more than 40% over 4 cores.and that was a dogfight with 20-30 planes visible and was bombing ground under them and also recording game video.running TS or discord.so overclocking a cpu that is not being maxed out will do little to no good to fps except if overclocking also FSB and PCI-E and memory (prob biggest boost)
dburne Posted June 30, 2017 Posted June 30, 2017 If youy are indeed running at 4.2-4.4 GHz on air, and stable, I would not try to squeeze any more out of it. You would not gain much at all.
SCG_Space_Ghost Posted June 30, 2017 Posted June 30, 2017 (edited) -snip- IL2 is maxing out your 4790K? NO WAY EVER!! -snip- Didn't see that anywhere in the OP. Fake news. Additionally, I wouldn't recommend a FSB OC and I'm not sure why you would either. -snip- As Don pointed out, 4.2-4.4GHz is a good OC for the chip and you're unlikely to see any significant gains past that. I don't see much of a gain past 4.4GHz on my i7 6700K but I still OC it because it runs at an acceptable voltage/temp for the speed. Edited June 30, 2017 by Space_Ghost 1
Jade_Monkey Posted June 30, 2017 Posted June 30, 2017 (edited) My only recommendation is that you run all cores are the same speed. Just check the box for "apply to all cores". For max OC, keep an eye on temps and just keep pushing for highest stable multiplier. Then once you figure that out, start reducing the vcore until its not stable. Then kick vcore up a couple of notches and you should be good. Thats an oversimplified quick guide. Edited June 30, 2017 by Jade_Monkey
Cloyd Posted June 30, 2017 Posted June 30, 2017 Just curious. chiliwil69 probably is not likely to see much improvement with overclocking. I have a 4770k running at default 3.5. Would it make an appreciable difference to OC? When I built this rig I got the K processor just to have the option to OC. When it gets old and is ready to be brought to the dump, I'll probably crank it up and see what the limits are. But, for now performance is acceptable and I don't want to turn up the heat. Is it worth stressing the processor for what might be just a few extra FPS? Cloyd
DD_Arthur Posted June 30, 2017 Posted June 30, 2017 In my experience, raising your clock speed from 3.5 to over 4ghz will make an appreciable and very worthwhile difference to BoS. If this can be done comfortably then it is definitely worth doing. Also, as the OP has realised, its better to do it manually rather than using auto settings. Oh yeah, please ignore post #2 in this thread unless you really, really want to blow your chip, mobo and graphics card . 1
SCG_Space_Ghost Posted June 30, 2017 Posted June 30, 2017 (edited) -snip- Oh yeah, please ignore post #2 in this thread unless you really, really want to blow your chip, mobo and graphics card . Public Service Announcement... For the love of all things techy... Do not take advise from Poster 2 regarding hardware. This is not personal. This is not an insult. It is meant for your safety and the safety of your computer hardware. Edited June 30, 2017 by Space_Ghost
SharpeXB Posted July 1, 2017 Posted July 1, 2017 Overclocking can indeed be a benefit in BoS. My old i7-3770K became unstable running at 4.7GHz and so I disabled that and went back to stock speed which I think was 3.5. Huge drop in performance (DX9 game version). I eventually replaced it.
=TBAS=Sshadow14 Posted July 1, 2017 Posted July 1, 2017 (edited) Public Service Announcement... For the love of all things techy... Do not take advise from Poster 2 regarding hardware. This is not personal. This is not an insult. It is meant for your safety and the safety of your computer hardware. Would you cut it out and stop Trolling my posts.. As you clearly don't know what you talking about. And yes it's personal as you keep trolling my posts Reported to Mods for Personal Attack / Trolling Been working with computers Longer than you been licking your monitor ********************************************************************* Anyways moving onto the OP Why overclock if you are not using the CPU? They did not overclock CPU's for FSX because they were getting 25fps on 40% CPU Usage. (like in IL2 for those of you struggling) You overclock when you are getting low frames and CPU usage pegged @ 99% and GPU not being used as its bottle necked. 4.0Ghz @ 20% Usage(IL2 on dx11) 4.4Ghz overclock @ 18% Usage (IL2 on dx11) Will net the Same FPS and performance ignoring ANY placebo effects. So Yes overclocking a CPU thats 3.5Ghz and pegged @ 99% usage will help if you pull it upto 4-4.5Ghz and its then its only on 75-80% Usage (now working correctly), but if its still pegged @ 99% bottlenecked you will notice little to no performance increase as STILL Bottlenecked. BUT with IL2 Intel and AMD's are not anywhere being maxed out. this game uses a lot of cpu but then still its only around 20-25% @ 4Ghz on a GOOD 4core cpu (Heck FX8350 is barely used so THE AWESOME WORLD WINNING INTELS CPU usage should be about 10-15% in very heavy Mutliplayer combat.) Edited July 1, 2017 by =TBAS=Sshadow14
Bearfoot Posted July 1, 2017 Posted July 1, 2017 So, I have an i4790K on a Z97 Mk I motherboard. I'm happy to post my settings, but, as I'm sure you know every chip is different! So, you really have to experiment yourself: tweak this, run stability tests, repeat, etc. Can take quite a while. You also have to strike the balance yourself, as well. Fact is, OC'ing is not difficult, but if you are asking for the best settings on a forum, you might not want to OC! That's what I was told, and having gone through the process, I now see why. You want to learn the process/technique, and apply it yourself. Settings are not transferrable, but techniques are. Over on the DCS forums, some great advice on approaches, tricks/tips, as well as what software to use for stress testing and how to use it by BitMaster and Demon_. From there, scour the OC forums. A simple google search on specifically OC'ing the i4790K will turn up a couple of really great how-to's, and following their steps made it clear that the "every chip is different" dictum holds so true: no way I could replicate their settings and get the same results. But I could replicate their trial-and-experimentation techniques to get great results. Annyway, I've now overclocked it to 4.7Mhz, using a vcore of 1.35. Really, really, really, stable. More importantly, for the long-term health of my chip, max wattage never exceeded 1.5 (or even 1.26) under the most hefty tests I threw at it. Temps were cool across the board --- so if temp alone was a concern, I probably could have upped the vcore quite substantially to hit 4.8 or even 4.9. Of course, I have a great cooler (Noctua NH D 15S). Here is my OC log of the latest configuration: - Date: 2017-05-12 Time: 2100 BLCK: 100 Core: 47 Effective_Frequency_Hz: 4700 Vcore: 1.325 CPU_Core_Voltage: Manual Load_Line_Calibration: 7 CPU_Power_Current_Capability_Percent: 130 CPU_Phase_Power_Control: Extreme CPU_Power_Duty_Control: Extreme DRAM_Power_Current_Capability_Percent: 130 DRAM_Phase_Power_Control: Extreme
coconut Posted July 1, 2017 Posted July 1, 2017 You overclock when you are getting low frames and CPU usage pegged @ 99% That's wrong. Your CPU can be the bottleneck even though usage is less than 99%. To take an analogy, a car can be too small to fit a sofa in it, even though the volume of the sofa is less than the volume of the trunk. For multithreaded software, an uneven distribution of work between threads can lead to the CPU being the bottleneck even though not all cores are fully used. The faster threads are done early and wait for the slower thread. The excess CPU resources unfortunately cannot be used to help the slower thread, because the slower thread can only execute on one core at any time. Before someone says "poorly optimized game, why don't the devs fix it": that's not easy to do. Games are computation-hungry programs with real-time constraints that have to react to user input as quickly as possible. Offline simulations, 3d/video renderings are examples of CPU-hungry programs that have no real-time constraints and whose entire input is known in advance. Those kinds of tasks can be more easily distributed across cores, leading to higher overall CPU usage. 1
chiliwili69 Posted July 1, 2017 Author Posted July 1, 2017 (edited) Well, to put some facts on the table I have been doing a series of tests with the "Balapan recorded flight" in a monitor. (https://forum.il2sturmovik.com/topic/29322-measuring-rig-performance-common-baseline/ ) I have using CPU clock speed from 4.0 to 4.7 GHz with increments of 0.1 GHz using "MSI Command Center" with the voltage in Auto (I am just a beginner). I tried 4.8 GHz but I had blue screen. It is surprisingly easy to modify clock speed with the MSICC without entering in the BIOS. For every clock speed I have done three passes of the recorded flight and recorded the actual fps with Fraps exported in a .csv file. Later I calculated the average of the three .csv files and produced the following summary chart: The low fps are when there is a lot of smoke in the scene. I also calculated the average fps for every speed and put it in a graph: There is a 6 fps increase when going from 4.0 to 4.1, and later the increase is about 2.5 fps for every 0.1 GHz. So going from 4.0 GHz to 4.7 GHz I gain 21 fps which is not bad at all. (and free! well some more Watts...) Therefore, I can empirically conclude, without fear of being wrong, that the OC speed influence the fps BOS performance for an i7-4790K in my rig. Now, I will perform a similar test but for VR. And, If the trend is similar, I really will need to squeeze my CPU as much as safely possible. Edited July 1, 2017 by chiliwili69 2
Jade_Monkey Posted July 1, 2017 Posted July 1, 2017 Does this game benefit from higher memory clocks?
dburne Posted July 1, 2017 Posted July 1, 2017 (edited) Does this game benefit from higher memory clocks? Maybe some, but certainly more in the CPU speed department. Getting back to chilliwill69 OP, running 4.2-4.4 I suggest he just not worry with it and have fun with the game! He would just not gain anything substantial by trying to achieve a little higher clock, he might get 4.5 doubt would get any higher. And with his chip would be easy enough to try by just setting it to manual and upping the CPU multiplier to get to 4.5. May not even need to make a voltage adjustment. Just set it, run say Prime 95 Blend and monitor cpu temps to check for stability. If Prime crashes just add a little vcore and test again. Being on air though I would keep a good eye on those core temps. I would also suggest whilst in the bios to disable Hyper Threading, helps the system stability, cores run a little cooler, and games really do not benefit from having Hyper Threading enabled. These K series of Intel processors are so dang easy to overclock. I built my current rig in Dec of 2013, and this thing has been running at 4.50 GHz daily ever since, and never had a single issue with it. I have been building rigs for more years than I care to remember, and this one is by far the best and most stable of all I have done. Edited July 1, 2017 by dburne
chiliwili69 Posted July 1, 2017 Author Posted July 1, 2017 (edited) I eventually replaced it. I have seen that you are running your 4790K at 4.7 GHz with Liquid cooling. Are you choosing a manual voltage? What is it? What temperatures do you have when running BOS? From previous post I believe you still not use VR but you play at 4K. So, have you tried to go beyond 4.7? Edited July 1, 2017 by chiliwili69
chiliwili69 Posted July 2, 2017 Author Posted July 2, 2017 (edited) I have conducted the same testing than above but with the Oculus Rift. The overclocking frequencies has been 4.0, 4.3 and 4.6 (three tests per speed) The settings of the Balapan test for VR are detailed in the another post indicated above. It is important to note that I put the HUD OFF (pressing H) before recording with Fraps, since the HUD decrease the performance. Also there were no supersampling. The ASW was not disabled using Oculus Tray Tool, but just pressing Ctrl+NumPad1 in the game before launching the test. Also, thanks to the "Don trick", the Oculus Home is not launched which is less annoying. The CPU is 4790K, the GPU GTX 1070, the RAM speed 1866MHz. The results of the 3x3 runs are: 4.0 GHz Frames: 3524 - Time: 60000ms - Avg: 58.733 - Min: 43 - Max: 91 Frames: 3315 - Time: 60000ms - Avg: 55.250 - Min: 43 - Max: 91 Frames: 3451 - Time: 60000ms - Avg: 57.517 - Min: 43 - Max: 91 4.3 GHz Frames: 4211 - Time: 60000ms - Avg: 70.183 - Min: 43 - Max: 91 Frames: 4067 - Time: 60000ms - Avg: 67.783 - Min: 43 - Max: 91 Frames: 4149 - Time: 60000ms - Avg: 69.150 - Min: 44 - Max: 91 4.6 GHz Frames: 4480 - Time: 60000ms - Avg: 74.667 - Min: 44 - Max: 91 Frames: 4510 - Time: 60000ms - Avg: 75.167 - Min: 44 - Max: 91 Frames: 4260 - Time: 60000ms - Avg: 71.000 - Min: 44 - Max: 91 Calculating the Average of the avg we have: 4.0GHz - Avg 57.16 4.3GHz - Avg 69.03 4.6GHz - Avg 73.611 The averages fps during the 1 minute are: We can see there is a measurable influence on fps. We can then calculate the gain ratio (fps per 0.1 GHz), this is from 4.0 to 4.3 we have went from 57.16 to 69, so the ratio is about 4 fps per 0.1 GHz. (Here the CPU was clearly the bottleneck) For the 4.3 to 4.6 the gain is 1.5 fps per 0.1 GHz (Here there seem to be other bottlenecks) For the 4.6 GHz the core temp were always below 75 Celsius using a Vcore voltage of 1.21 Taking into account that the Balapan test is quite demanding, in a simpler scenario the fps should be at 90 fps, which is what I wanted after all this CPU upgrade (from 4790 to 4790K). Thank you all for your wise input. Edited July 17, 2017 by chiliwili69
chiliwili69 Posted July 2, 2017 Author Posted July 2, 2017 I would also suggest whilst in the bios to disable Hyper Threading, helps the system stability, cores run a little cooler, and games really do not benefit from having Hyper Threading enabled. Yes thanks, HyperThreading was disabled for all tests from the BIOS
chiliwili69 Posted July 2, 2017 Author Posted July 2, 2017 I've now overclocked it to 4.7Mhz, using a vcore of 1.35. Thanks for your input. You said vcore 1.35 but below you write 1.325? Yes, every chip is different when OCs, so those values are just for reference. As I see, this OCs limits are constrained by the amount of heat you can extract from the chip. So, it is very important a CPU good cooling when budgetting a new rig, at the end is going to be you limiting factor. I have spent 285€ for a new i7-4790K (it was a bargain, a shop was closing and were selling stocks) and in one day I sold my 2.5 year-old i7-4790 in ebay for 185€. The Sycthe Katana-4 CPU air cooler I was buying as well (30€), was the best cooler for the available space of my micro-ATX boxcase. Total investment of the upgrade 130€. But, in my future rig, (1.5 to 2 years from now) I will be looking for a big boxcase where I can put a big CPU cooler (air or perhaps liquid) to be able to meet the future requirements of 2nd gen VR, that for sure will be quite demanding.
Bearfoot Posted July 2, 2017 Posted July 2, 2017 Thanks for your input. You said vcore 1.35 but below you write 1.325? Yes, every chip is different when OCs, so those values are just for reference. As I see, this OCs limits are constrained by the amount of heat you can extract from the chip. So, it is very important a CPU good cooling when budgetting a new rig, at the end is going to be you limiting factor. I have spent 285€ for a new i7-4790K (it was a bargain, a shop was closing and were selling stocks) and in one day I sold my 2.5 year-old i7-4790 in ebay for 185€. The Sycthe Katana-4 CPU air cooler I was buying as well (30€), was the best cooler for the available space of my micro-ATX boxcase. Total investment of the upgrade 130€. But, in my future rig, (1.5 to 2 years from now) I will be looking for a big boxcase where I can put a big CPU cooler (air or perhaps liquid) to be able to meet the future requirements of 2nd gen VR, that for sure will be quite demanding. Yes, sorry -- I meant 1.325. And temps are, of course, crucially important for immediately not frying your machine, but my understanding is that wattage above 150 or so can kill your machine just as surely over the longer term. My point was that with the Noctua air cooler and 1.325 vcore, the temperature was simply not a concern --- lots of headroom --- so when I popped into high vcores where I could get a stable 4.8, the temps where still manageable, it was the wattage that was the barrier. The Noctua air cooler, BTW, is excellent!
SharpeXB Posted July 3, 2017 Posted July 3, 2017 I have seen that you are running your 4790K at 4.7 GHz with Liquid cooling. Are you choosing a manual voltage? What is it? What temperatures do you have when running BOS? From previous post I believe you still not use VR but you play at 4K. So, have you tried to go beyond 4.7? The overclocking was done by my PC vendor, I'm really not knowledgeable about the details. I'm just running a monitor, no VR
Dutchvdm Posted July 3, 2017 Posted July 3, 2017 Mine is running nicely at 4,7 Ghz. Before the DX11 patch this was a huge upgrade from the normal 4.0/4.4 modus. Now i don't know.. The sim might be more GPU hungry with DX11, but at least it won't hurt either. I'm running it with a different non stock cooler. No water cooling here. The later 47xx... should perform better than the early models. Mine's from end of 2015. Grt M
A_radek Posted July 3, 2017 Posted July 3, 2017 Does this game benefit from higher memory clocks? Yes, in my experience and to my own suprise it does. so if you bought ram faster than default ddr4 2133mhz, and don't feel like overcklocking anything. At least go in to your bios and enable XMP. That's all here is to it. It's a factory profile built in to your ram, that tells your bios at what speed, voltage and timings to run your ram at. Many seem to buy fast ram without ever doing this, meaning they are running their unnecessarily expensive ram kit at default ddr4 2133mhz speeds. Did these tests on a regular monitor, running a 77000k cpu and a 1070 gpu. @4.2ghz and ram at default 2133mhz: Frames, Time (ms), Min, Max, Avg fps 8637, 60000, 114, 186, 143.950 @4.2ghz and ram at 3000mhz: Frames, Time (ms), Min, Max, Avg fps 9184, 60000, 118, 200, 153.067 @4.8ghz and ram at 2133mhz: Frames, Time (ms), Min, Max, Avg fps 9266, 60000, 120, 198, 154.433 @4.8ghz and ram at 3000mhz: Frames, Time (ms), Min, Max, Avg fps 9869, 60000, 124, 209, 164.483 Found these results so suprising I did another one. @4.5ghz and ram at 3000mhz: Frames, Time (ms), Min, Max, Avg fps 9637, 60000, 120, 207, 160.617 In my case, enabling xmp gave a framerate boost equivalent to a 4-500mhz cpu overclock.
chiliwili69 Posted October 29, 2017 Author Posted October 29, 2017 (edited) Well, after upgrading my rig to a proper ATX case (it was micro-ATX) and putting a NZXT Kraken x-52 AIO liquid cooling I have been experimenting with higher OC and stable Vcores. There is a nice page with many OC tests of 4790K users: http://www.overclock.net/t/1490324/the-intel-devils-canyon-owners-club I have downloaded that table, eliminate some nonsense fake test and plot them for every freq and Vcore. Then I tried several freq and manual Vcore and check stability with Prime95 with blend torture test with one or two threads (which is much more demanding that BOS in VR). These are the my stable Vcore for very freq: CPU GHz Vcore 4.0 1.071 4.4 1.156 4.6 1.215 4.7 1.249 4.8 1.293 4.9 1.342 5.0 1.381 Regarding temperatures, the kraken X-52 manage quite well the Prime95 stresstest, being around 80C (peaks to 85) for the 5.0GHz With my previous microATX case and mid-range CPU aircooler I was playing BOS at 4.6 Ghz. Now I will experiment with these higher freqs in BOS VR. Just for the records, this is the procedure I followed for CPU OC: 1. Launch MSI afterburner to trend CPU temp, Cores temp, CPU clock, etc, etc. (you can put sample interval 0.1 sec) 2. Set CPU Vcore voltage to Man. (for example 1.15 v) (using MSI Command centre in my case) 3. Set freq to all cores to 4.2 GHz (using MSI Command centre in my case) 4. Run Prime95 Blend torture test with one thread. 5. Leave that for 30 minuntes and monitorize CPU temp and Core temps. If you have BSOD, then you should increase the voltage a bit (for example from 1.15 to 1.17) If you don´t have BSOD, then you should decrease the voltage a bit (for example from 1.15 to 1.13) It is better that you write down in a paper every test (anotate CPU GHz, Vcore voltage, CPU temp), at the end of the procedure you should keep the minimum stable voltage. Edited January 25, 2018 by chiliwili69 1
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now