Intel klar med opdatering til Haswell-serien

Karstenkranvogn (4) skrev:

Det ser ud til de nye HR modeller blot har en clock forskel på 100MHz højere end fra tidligere Haswell modeller. Og tilsynladende er der ikke nogle K modeller i de nye HR modeller. Betyder det så, at man ikke ville kunne overclocke nogle af CPU'er fra den nye HR serie? Og siden der kommer en Haswell Refresh serie nu, betyder det så at Broadwell er blevet udskudt?

Karstenkranvogn (4) skrev:

Det ser ud til de nye HR modeller blot har en clock forskel på 100MHz højere end fra tidligere Haswell modeller. Og tilsynladende er der ikke nogle K modeller i de nye HR modeller. Betyder det så, at man ikke ville kunne overclocke nogle af CPU'er fra den nye HR serie? Og siden der kommer en Haswell Refresh serie nu, betyder det så at Broadwell er blevet udskudt?

Montago.NET (3) skrev:

men du har lidt ret.. Til gaming er der ikke megen forskel på en Sandy, Ivy eller Haswell

http://www.tomshardware.co.uk/forum/348248-28-overclock-processor skrev:

Correct me if I'm wrong, but the reason .1 Ghz seems like a lot on a GPU and not on a CPU is sheer relativity. .1 out of 3.3 only equates to an extra 3% cycles per second, while 100 Mhz out of a 1024 Mhz GPU is a full 9.8% extra clocks per second. Simply put, because GPU's are generally run at slower clocks than CPU's, each Mhz of OC makes a bigger difference in GPU's than it would in a much faster running CPU.

To use (slightly) more complicated math:

3,300*60 seconds = 198,000 cycles per minute

3,400*60 seconds = 204,000 cycles per minute

GPU

1024*60 = 61,440 cpm

1124*60 = 67,440 cpm

So with a CPU you get 6,000 more cycles per minute, while on a GPU you get 6,000 more cpm as well, so theoretically, the OC is the exact same. But the GPU will seem much faster, an almost 10% increase, simply because 6,000 cpm is a lot to add to 60,000, while it isn't so much to add to 198,000.

That ends my simpleton discussion of relevancy.

Montago.NET (16) skrev:

Jeg fandt svaret på hvorfor non-K CPU'er er svære at OC, umiddelbart er det iGPU som ikke kan tåle det.

http://www.tomshardware.co.uk/forum/348248-28-overclock-processor skrev:

Correct me if I'm wrong, but the reason .1 Ghz seems like a lot on a GPU and not on a CPU is sheer relativity. .1 out of 3.3 only equates to an extra 3% cycles per second, while 100 Mhz out of a 1024 Mhz GPU is a full 9.8% extra clocks per second. Simply put, because GPU's are generally run at slower clocks than CPU's, each Mhz of OC makes a bigger difference in GPU's than it would in a much faster running CPU.

To use (slightly) more complicated math:

3,300*60 seconds = 198,000 cycles per minute

3,400*60 seconds = 204,000 cycles per minute

GPU

1024*60 = 61,440 cpm

1124*60 = 67,440 cpm

So with a CPU you get 6,000 more cycles per minute, while on a GPU you get 6,000 more cpm as well, so theoretically, the OC is the exact same. But the GPU will seem much faster, an almost 10% increase, simply because 6,000 cpm is a lot to add to 60,000, while it isn't so much to add to 198,000.

That ends my simpleton discussion of relevancy.