Free performance, why not? Actually yes I can tell the difference between 55 and 60 fps, I know as soon as I am not locked at 60!
At stock speeds pretty much all of the Sandybridge range were really lame. If you take a look at this, for example. http://ark.intel.com/products/52207/Intel-Core-i5-2400-Processor-6M-Cache-up-to-3_40-GHz I had one of those. The fastest it would boost to was 3.4ghz or so. When I compared it to the results people were getting from 2500ks at around 4.7ghz I really regretted it. Sandybridge only really came into its own due to the extremely lame clock speeds. So yeah, I bought a 3970x. http://ark.intel.com/products/70845...ssor-Extreme-Edition-15M-Cache-up-to-4_00-GHz Which again, lame. It says it boosts to 4ghz but that's only on one core. On all of them IIRC the boost clock was around 3.8ghz, with it being a hex core CPU and all. Minimum FPS differences with the CPU at stock compared to the overclock of 4.9ghz were enormous. In Cinebench I could just about break 1300 points. That's with the CPU at a not completely stable 5ghz. With it at stock? around 957 points. So that's about a 30% speed boost. At first I was gaming at 1080p with a cheap 3D monitor I bought until the prices of 4k screens came down. I would say that even at 1080p I was CPU limited (two Titan Black). The FPS were insane of course, but the minimums were far better with the CPU overclocked. I would say 25% which is a lot on a minimum. When I went to 4k I just left the CPU as it was. However, when it died I replaced it with a 2.1ghz 8 core Ivy chip and in certain games like GTAV the minimum FPS were exactly the same. Crazy, considering it was over twice as slow clock speed wise but I guess that's 4k for you ! everything is down to the GPUs. So yeah, Sandybridge were massively faster when overclocked. Due to Sandy clocking better than Ivy or Haswell I was able to beat the 3930k, 4930k and 5820k with it running at 4.9ghz. Sadly it seems things are just not made to last. So much for MSI's 24 phase (or might have even been higher, this board had stupid power phases !) Military grade components. They lasted for 11 months. So then I RMA the board and they email me 6 weeks later to say they don't carry any stock of X79 boards. That really, really upset me which is why I'm very careful now. That was a £3500 rig and it was rendered useless because I could not get a replacement board for it. In the end I found a second hand RIVE on Ebay before the prices went stupid and paid £125 for it. I quickly sold it on.
Part of overclocking for me was always the fun of doing it, Gains in fps or performance where never my goals when overclocking. Was just fun to do, It always felt like after they made pc building like Lego it was the last bit of difference a user made pc could do compared to a pre build. I could of brought my spec with warranty and pre overclocked for £150 less than I have actually paid to do it myself. PC building and overclocking to me has always been the fun of doing it msyelf. The day I stop overclocking is the day I buy a pre build. Its still a expensive hobby to me and always probably will be, Its a similar reason I got into RC Cars / Helicopters and planes. Not to race them but to actually build from scratch the entire thing. Yeah you can buy them fully assembled but wheres the fun in that.
Perhaps a small overclock of my I5 6600K then. Anyone care to point me in the directions of how to do this safely?
Personally it's not about gaming - but instead there's some significant (100s of GBs at a time) audio multithreaded batch processes & video editing tasks where it makes a material difference to the process time. in the former case this was having (for years now) removed a single 6Gb/s SSD as being the bottleneck - 2 in R0 still being fast enough for an o/c'd 5960X. is this with a standard monitor or using g-sync/freesync? Yeah, just wondering if it's the card or the monitor that makes it noticeable to you - as unless things were dreadful, i personally didn't notice any major difference with fps when gaming for over a decade on old 2048x1536 CRT monitors or (subsequently) with a g-sync? Otherwise, obviously the %age difference is greater between 55 & 60fps vs ~85 (my approximation) & 90... ...& i should have taken this slightly more into account.
Actually I just sat down and thought about it and I got the 3970x two years ago, not last year. Last year I bought my Area 51 Triad machine. I have spent the time (it was good fun ) seeing what my rig can do. I mostly just keep it in reserve now because I am using software for both so I have saved settings I can apply in seconds if I need to run a bench or something Thankfully Alienware hold spare parts for years and years. That's why I decided on the pre built machine.
Ever helpful Parge. I was way ahead anyway and am following this. https://www.youtube.com/watch?v=pAr4VygoWWg
5820k costs £330 - the 5930k costs £500. The difference between them, aside from some PCIe lanes that I can't use (ITX), is 100mhz in clockspeed. So, I can raise the multiplier by 1x and save myself £170..... hell, I could probably raise it 4x without changing anything else and hammer a stock 5930k. Takes as long as it takes to boot into UEFI and restart. So, surely a no brainer?
Graphics card I wouldn't bother unless the model I have is known to have just way too much headroom to ignore. On the CPU front I can't think of any Intel ones in recent memory where you couldn't get an extra 500-1000 Mhz with little to no hassle. I have some games that are very sensitive to single core performance so for me that's usually a perceptible difference in quality of life.
Except you could also raise the multiplier on the 5930k by 4x and you're back to square one in terms of a performance delta.
I know you're probably aiming this at people with more recent CPUs than I have, but it's made a huge difference for me. Like I said earlier, the older/lower spec your hardware, the more there is to gain by overclocking it. For example, when the 2500k was released, there probably wasn't a game available that it couldn't handle at stock speeds. Fast forward 5(?) years and the 2500k is still a fantastic chip, but only if overclocked. My ancient X3360 (Q9550 C2Q) is further down that same path so needs the overclock to even get a look-in. At stock (2.8GHz) it simply wouldn't have a chance, but at 4GHz it keeps 64 player multiplayer Battlefield 4 (1080p, 4xmsaa, no fxaa, everything else maxed) almost locked at 66fps (vsync on and monitor refresh overclocked), with MSI Afterburer showing all 4 cores hovering around 90-95%+ usage and maxing at 100%. She's giving it all she's got, kiptin.
Looking at that video the voltage seams high in the tests. for a 6700k I use 46 multi, 1.32v adaptive. I normally start at 45 multi and 1.35v and test from there, see if the cpu will do 46 and then see if it will be fine with slightly lower volts. The settings should be very similar for a 6600k For stress test, I use intel burn test and watch the temps on real temp.
Is it really that clear cut though? On average a higher end version of a CPU will probably be better binned, but I'm not so sure that an average mid ranged one will necessarily be that much weaker quality.
No its never that clear cut. If nothing else, the higher CPU will have a higher clock and the same multiplier value will have a greater impact on speed than the same multiplier applied to lower end CPU. Comparisons become less clear cut when overclocks become involved. My point was that comparing an overclocked component X to a non overclocked component X+1 doesn't make much sense when you can also overclock X+1. Even though the results will vary from component to componet you should really compare them both overclocked or both at stock.
