Hello (I know about the 2500k...cant buy it neither need it) I saw a i5 2400 at a store today at a very good price and it seems like a good upgrade. Since i dont overclock like hell and as i pretend to keep my cooler master hyper 212+ with good temps, this seems a good choice. My question: can the 2400 be overclocked at the same frequency in all cores? I mean, put the max multiplier, which is 38x, and have all cores running at 3800mhz under load(100 base clock)? I see a fuse all over the internet regarding these non k cpus saying that it will be less and less the overclock wheter your´re using 1 core(3.8ghz), 2 cores(3.7ghz), 3 cores(3.6ghz) etc By the way, safe base clock? Guess i should not touch that but is it safe to set it to like 103-106mhz? That´s all. Thaks in advance
that cooler is good for about 4.5ghz on a 2500k. I would not mess with baseclock oc on 1155 too much is tied to it.
Thanks for your answer. I know both things you said. But that dont answer my main question really...:nerd: The i5 2400 is affordable for me atm, the 2500k not and i dont really care for such a high overclock. Thanks again
You can't over clock non-K chips for Sandy Bridge based CPU. If you over clock without using the unlocked multiplier, you will burn other components on your computer. Deltatux
Turbo frequency scales based on core usage....so, 1 core = 3.8ghz, 2/3 cores = 3.7, 4 cores = 3.6ghz. That's how it's designed to function. It's not advised to adjust the bclk at all. Overclocking is done using the Turbo multiplier....which allows 4 bins over stock on non-K chips.
The Core i5 2400 is 190$, the core i5 2500K is as low as 200$ some places. What is the price you see the 2400 at? Bit Tech says this though:
99% sure you´re wrong. It´s not unlocked, but has a high limit, depeding on the cpu: The 2300 has max 35 multiplier. The 2400 has max 38 multiplier. The 2500 has max 41 multiplier. Right?
Because i was not sure about my main question, which was if all cores would raise up to max multiplier under load. Thank you.
Correct And the correct answer is no, not by design. Some sites claim that you can enforce the turbo multiplier by disabling Turbo and SpeedStep though. I haven't tried it personally as my system performance to my liking....however, if you'd like, I'll give disabling those settings a shot and we'll see what happens.
Thanks sykozis for the answer. I would be very grateful if you can test that thing and tell me the result.
Will do....just give me a few minutes to change the settings and I'll update this post with the results. Update 1: Disabling SpeedStep was pointless.... Update 2: Disabling C1E and SpeedStep was also pointless.... I don't have an option to disable Turbo Boost, so I can't try that, but thinking about it....if you're overclocking using the Turbo multiplier, it stands to reason that disabling Turbo Boost would effectively disable any type of overclock....
Thank you, it was very kind of you. So, it works as you described in your first post - 3.6ghz all cores. Good enough for me, considering the shop is only 5km from my house and its the best price in the country atm - 134.99€. They´re clearing the stock i guess, i must do my part for the community!:nerd: Give me the wellcome to the club:wanker: Thanks.
It's an interesting question. According to Intel's examples you could overclock a 2400 to 3.8 with 1 core, but only 3.5 with all 4 cores active - ie +4 bins (where 1 bin equals +1 multiplier) from the stock numbers. Take a look at the graphics on these two pages from teh anandtech review: http://www.anandtech.com/show/4083/...-core-i7-2600k-i5-2500k-core-i3-2100-tested/3 http://www.anandtech.com/show/4083/...-core-i7-2600k-i5-2500k-core-i3-2100-tested/2 What I'm not sure about is whether you are forced to do a "stepped" overclock using individual multipliers for each core, or whether you can set all 4 cores to the maximum multiplier of x38. There's no reason the chip wouldn't support that, just whether or not the motherboard will limit the "4 cores active" multiplier ratio to x35 and only allow the single core multiplier to be set at x38. I don't think there's any way to really test it without having a non-K series chip to play with. To the OP - would you still be happy if 3.5Ghz was the highest you could go on all 4 cores?
So Sykozis, I didn't realize you were actually running a 2400, were you able to set x38 across the board, or just a single core? ie, what does CPU-Z say for clock freq when you run prime 95?
Nope...only a single core. 1 core = 3.8ghz 2/3 cores = 3.7ghz 4 cores = 3.6ghz Anandtech is partly wrong....btw.... Intel's Turbo Boost isn't designed to allow different multipliers to be set for each core. It's specifically designed to scale the clocks speeds down according to core utilization. 1 thread = 3.8ghz 2/3 threads = 3.7ghz 4 threads = 3.6ghz That's as reported by RealTemp. When running Prime95, I close everything but Prime95 and Realtemp. Also, 3.6ghz is reported by RealTemp when running Folding@Home SMP, which applies 100% load to all 4 cores. I could always go back to my i7 870....but this i5 2400 outperforms it rather well, while drawing less power. In Folding@Home SMP, this i5 2400 gets the same PPD as my i7 870 with HT enabled. Plus it runs cooler....
Thanks for the answer Sykozis. I didn't mean to suggest that you could set a multiplier for each core by itself, but that many mobos (Such as my p8z68 deluxe) offer a per core multiplier setting. I can set a different multiplier for when 1,2,3 or 4 cores are used. Right now I set a x44 multi on all 4 settings, but if I wanted I could set 4 core usage at 44, then have the system turbo up to 45, 46, and 47 as less cores are needed and can be power gated. There's a guide on Hardocp here: http://hardforum.com/showthread.php?t=1592713 Kind of temping to try it out, but I don't really want to have to crank the voltage up from 1.256 where is is now. I have an i7 860 in an htcp downstairs that I normally use for media encoding but like you said, even without hyper threading the sandy bridge cpus really outperform lynnfield. I've been meaning to do some tests to get some numbers so I can see how much by.
The performance difference seems to be a bit situational. In WoW for example....I had to enable v-sync because my framerates in 1 particular area that I visit most frequently went from 200-250fps....up to 380-430fps. In 3DMark11, I gained around 100pts. In Folding@Home SMP, my PDD remained roughly the same. Not the best testing to determine performance difference....but WoW and Folding@Home are the 2 most frequently used apps on my computer.
I haven't really looked at gaming performance as I only have a 5830 gpu downstairs ( i like to play racing games on the bigger screen), but it wouldn't be a fair comparison. I use handbrake for media encoding (which does not utilize the gpu) and it shows the fps as it encodes, but I need to see if I can get it to output a log at the end of an encode to show the final average performance. Then I can have both machines rip the same video and compare the results.
