Well, I'm OK with it because the only moment I even could see the 10 cars is at the start of the race. It doesn't matter if can see 10 or 58 cars during the race because there rarely is that even competition.
I tested CPU Z... what can you say about this: View attachment 5876 Core speed went to 3693 when I loaded a server (hosting). So is this the perfomance I should have with my CPU?
That would be CPU auto throttling the clock speed down to be more power efficient during periods of idleness. I presume you weren't running any cpu intensive program when you took that picture? It's set in the bios and should be called "C1E Enhanced Halt State".
Nothing special in background. I am ok with it if the speed goes down ONLY when idle. My CPU should have this "turbo mode" which should give me 4,0 ghz instead of normal 3,4ghz...do you know if this is supposed to activate itself when the CPU is in heavy use?
This is the first time I have heard about turning off HT to give dramatic performance increases?? I know that its been recommended to turn it off for more stable OC's and I have done that in the past as I didnt require the additional cores on my gaming pc. I still have not heard anything about 'significant' increases in performance with it off. Not sure I agree with your maths though, 2/8ths is the same as 2/4 when in the end the games is still getting 2 cores at a specific clock rate. It's not about percentages of cores, its about how much CPU you are getting - in both cases you are getting 2 cores. The only thing I can think of is that you are getting one core and one hyper core which I assume is not as efficient as the game running on two real cores. I also assume that this is not happening as the you can tell the OS that you dont want something running on a logical core.
I don't have an AMD cpu so i'm not sure but if i had to guess i'd say the amount of turbo added is automatic and varies depending on cpu usage measured via a built-in cpu or motherboard controller that monitors utilisation. I haven't switched on HT for at least a few years but from what i remembered, turning HT on artifically devides each core into two cores with an even 50:50 re-distribution of a single core's net potential. If a game can only use 2 cores by design (but any 2 cores it so chooses), then the total processing power of two cores on a HT cpu should effectively be the same as only 1 core on a non HT cpu. For example, if each core can do 2 processes per second, on a quad core that means 8 processes per second in total. If you enabled HT, then now the 4 cores become 8 cores but each core can only do 1 process per second. The math remains the same, 8 cores at 1 process/second = 8 processes/second total. And so any program designed to take only 2 cores will only get 2 cores at 1 process/second = 2 processes/second total on a HT cpu vs 2 cores at 2 processes/second = 4 processes/second total on a non HT cpu. The only time HT is beneficial is if a particular case when using a program satisfies two conditions: 1) The program can scale to use as many cores as there are available. (jobs such as 3dsmax rendering do this very well because each job handed to a core is non-reliant/independent of the results from jobs being carried out by other cores). 2) If the task given to a core is not fully utilising the full potential of that core. This happens when a particular task given to each core is being completed too easily/fast by a core therefore resulting in some wasted processing potential.
I`d rather be with the same FPS ...but better textures. The collor track seems like this pan: View attachment 5877 But with rubber on real road. Fells like no texture.
Does anyone have benchmark numbers in rF2 or rF1 to prove this? Unfortunately my HP dv7 laptop bios is crippled and does not allow disabling hyperthreading, otherwise I would try.
That is not how it works, if it was then single threaded applications would run at 50% of what they are capable of if HT was turned on. Of course they dont they still run at full speed. It doesnt make sense that you double your cores and your actual computing capacity does not increase. If thats how it behaved then what would be the point? We may as well do what we have always done and used software threads managed by the OS. They always worked fine. What you describe is not how HT really works and with the OS's knowledge of the cpu topology it would be strange that it would schedule two intensive threads on the same core. It's also inconsistent with the example that 3dsmax can give better performance if in fact each thread has only 50% of a cpu's full potential. If that was the case it would run the same times.
For starters you can't gain more potential energy (i.e. cpu power) for nothing, but you can better utilise what potential energy that's at your disposal. Below is a working example when these following conditions are met.... ...example: If a program only assigns one task per core and for this example lets say only 50% of each core is being utilised/taxed on a non HT cpu, then there is a waste of 50% processing power per core. This amounts to 50% of the total cpu power being unutilised/wasted. If you then switched on HT, you would now have twice as many cores but each core only has 50% of the total power of a single core of a non HT cpu. Now, when you assign that same task per core (as with the non HT cpu), now that 50% utilisation = 100% utilisation per core on the HT cpu. Let me demonstrate with a working example.... If a Non HT cpu performs 2 processes/sec per core and each core is only utilised 50%, then the task is using only 1 process/sec per core with the other 1 process/sec being wasted doing nothing. When you turn HT on, each core can now only perform 1 process/sec and since the task only uses 1 process/sec then the core is being 100% utilised. Since the 4 cores are turned into 8 cores, you now have 8 cores running at 100% each whereas before without HT enabled you had 4 cores running at 50% each. That's double the cpu power being utilised by that program. The whole idea of HT is to increase the total utilisation of a cpu's potential power for certain tasks where and when possible. Now you might ask why doesn't the program just assign two tasks per core (called multi-threading) so that 50% task 1 + 50% task 2 = 100% core utilisation. That's a fair question but the reality is that multi-threading whilst extremely good for utilising the most out of your cpu's potential, it is not always the case depending on the type of task(s) or sometimes if/when a developer does a bad job of it or misses it out all together. You can read more about the positive and negatives of multi-threading here to better understand why Hyper-threading originally came about: http://en.wikipedia.org/wiki/Multithreading_(computer_architecture)#Overview Now if anyone did have HT enabled and found that they didn't gain extra performance in game, i would guess the reason to be that their cpu's are running at a high clock speed (thus high processes/sec potential) and without HT enabled each core was less than (or dancing around the) 50% utilisation which would translate to no more than 100% per core on a HT cpu which would yield equal performance. The reality is that nowadays HT probably does more harm than good when the cpu is heavily taxed since developers design their software to be extremely well multi-threaded. For example, a game that only needs a 2Ghz CPU with no more than 2 cores. If the PC has a dual core 4Ghz cpu then each core will only be 50% used and the pc has twice the amount of power than necessary. If you then turned on HT, the cpu has 4 cores each running at an equivalent processing power of a 2Ghz cpu. Still only 2 cores are used but each to 100% and not a penny over hence the game performances will appear the same. Now before anyone tells me how can this be true when they only have (say) a modern 3.2 ghz cpu with HT enabled and rf2 has a recommended system requirement of a 3.0 GHz core 2 duo? The answer is that it depends on the cpu you have since performance is not just the core clock but also other specs of the cpu such as cache size, etc. Each new cpu generation performs better than the previous despite being set to the same clock speed. For example, i calculated that in order to perform as well a stock clock i5-2500k at 3.3Ghz my own quad core i7 920 requires an overclock in excess of 4.5Ghz which is simply unachievable on air let alone the best water cooling running stable. And to match an easily overclocked i5-2500k @ 4.5Ghz on air would require an insane 6+ Ghz i7-920 which is simply stupidly impossible. The recommended requirements for rf2 is an ever ageing core 2 duo which is something like the equivalence to a massively down-clocked i5-2500k @ 1Ghz (lol) if i had to make a guess.
Dumb question but I have a big monitor (42") and aliasing is very harsh which may be obvious but it's not obvious b/c pCars, for example, doesn't have this problem. So is bad aliasing something that is caused by bad graphics coding or does pCars have anti-aliasing built in or is it something else?
I highly doubt it's a coding issue. It's true that some game rendering engines can handle anti-aliasing badly but this has not been my own experience with rf2. Can you give some more details on what you've tried so far to resolve it? Every type of anti-aliasing level in rf2? Hit the reset to default settings everything in the nvidia control panel if you have an nvidia card or the equivalent on CCC? And also the same for the independent specific program tab? What card do you have? I have found this which may be of conformational use to you if it's the same sort of card you have: http://isiforums.net/f/showthread.php/2485-Antialiasing-doesn-t-work-correctly Also, can you take a picture with a camera of this aliasing problem?
DrR1ipper, you have got it completely wrong. You keep implying by your mathematical examples that HT halves the throughput of any core, this is not the case. The OS is managing the load across the cores. On the worst side of the equation the OS may schedule different threads on the same core - one to the physical processor and another to the same cores logical processor. That would be bad and would certainly see degraded performance. Two threads of high CPU usage should not normally be allocated like this when other cores are idle. In reality windows will put RF2 main thread on one CPU and the other main thread on another CPU, not on a logical core belonging to a cpu that is heavily loaded. Your assumption how things work is leading you to a conclusion that is wrong and not one that I have ever seen backed up by any benchmarks. Also, of your two examples up top. Number two is usually valid but number one is often an example of where real cores will give you benefit but logical cores do not, depending on the optimisation of the program. I am not expert in hyperthreading at the hardware level but I do have over 30 years with computers and multiuser/multthreaded systems from design/development perspective. I dont mind that you show me I am wrong but while you are showing me that you know how to add and subtract none of that information has shown me that this is how HT actually works. IF you have some references that show me thats how it all behaves then fine. Also, I am not saying in some rare cases it could not do something along the lines of what you say but it should not be the norm and is not something that should be regularly if ever an issue for RF2 players. I think somewhere along the line HT has been a must turn off for overclockers, for often valid reasons, and has since become something to blame for other things.
You can read about HT here: http://en.wikipedia.org/wiki/Hyper-threading some points of interest... HT halves the cpu cycle to each logical cpu independent of OS, meaning a single thread that requires 60% of logical cpu when HT'ed will mean that thread looses 20% processing power. http://www.makeuseof.com/tag/hyperthreading-technology-explained/ I'm sorry, i must make a correction. The above is only true when each core is totally taxed and each logical core is tasked with a totally independent task to the other so that two logical cores made from the same physical core are fighting for the same resources. If that is not the case and the two logical cores were paired to complete identical tasks requiring the same resources then you can technically gain a little more by than 50% potential from each logical core but never double. This is why tasks like 3d rendering applications reliant on cpu are so good with HT, especially when a single thread doesn't even use up the whole of a single physical core.
Jeez this thread should be retitled "In and outs of hyperthread". Regarding "optimization"... It's been discussed by us users in many threads but I don't think anyone from ISI (Tim) has ever mentioned a word about it. As far as I know it is a fantasy. If it actually happens I have very little hope that it will make the kind of difference I would need to see to make me happy. It's my number one wish but I'm prepared for disappointment.
Yes, I finally agree with the last part where two are saturating the cores then they could both be getting around 50% of the processor. What you have implied previously is that nothing can be more than 50% of the processor when HT is turned on. Now, in the context of RF2, things are more positive because we don't have enough going to to saturate cores all physical cores because RF2 does not use them. The items you underlined are also a little out of context because I assume that people are using a modern OS, e.g. windows 7. The example you gave is of an OS that is not aware of HT, in that case what you say is happening could happen, by chance, but on an OS that we use to run RF2 the underlined text is not applicable. On the second piece that you underlined, I very well remember that bottleneck that really hurt performance with HT when it first came out but I think the underlined text is more to do with the first offerings of HT, our core i5's i7's etc have advanced since then. What you quoted is based on processors nearly a decade old. DrR1pper, I very much get the impression that you are trying to back up what you have said by doing a bit of googling but still not understanding why it doesnt work how you are describing it.
