1 |
On 23/01/13 19:16, Volker Armin Hemmann wrote: |
2 |
> Am 23.01.2013 16:35, schrieb Nikos Chantziaras: |
3 |
>> On 23/01/13 17:09, Nilesh Govindrajan wrote: |
4 |
>>> On Wednesday 23 January 2013 07:52:03 PM IST, Nikos Chantziaras wrote: |
5 |
>>>> [...] |
6 |
>>>> In my experience, most of the time you can overclock. The issue is |
7 |
>>>> with the user not knowing exactly how to do it. You need to |
8 |
>>>> understand a few things and how they affect each other. It's not just |
9 |
>>>> a knob you can turn. |
10 |
>>> |
11 |
>>> That pretty much applies to me. I don't know much about hardware stuff. |
12 |
>>> Regarding your 1 Ghz overclock, you probably have good components in |
13 |
>>> terms of RAM & SMPS. |
14 |
>>> When I bought this rig in 2008, I knew nothing about good components, |
15 |
>>> blindly trusted local vendor... also internet shopping wasn't advanced |
16 |
>>> here. |
17 |
>>> So pretty much substandard components. |
18 |
>> |
19 |
>> The part that's really important is the mainboard. RAM doesn't |
20 |
>> matter. In my case, I had pretty basic 800MHz DDR2 RAM. Raising the |
21 |
>> FSB would bring it above that, so I changed the DRAM ratio to 1:1, and |
22 |
>> the RAM then ran at only 600Mhz. |
23 |
>> |
24 |
>> That was the starting point to rule out RAM problems. After that, I |
25 |
>> raised FSB but kept the VCore constant until I hit the first |
26 |
>> instabilities. When that happened, I raised VCore a bit. Rinse and |
27 |
>> repeat, until the VCore was still below the maximum recommendation by |
28 |
>> Intel. That happened at 3.4GHz (378MHz FSB * 9 CPU multiplier = |
29 |
>> 3402MHz CPU clock.) The E6600 CPU I got was an average sample. |
30 |
>> Others were running it at 3.6GHz (or even higher with water cooling.) |
31 |
>> |
32 |
>> This was a process that took about 3 days to complete (needs a lot of |
33 |
>> stability testing.) The good thing about those older CPUs was that |
34 |
>> the performance boost I got by OCing wasn't just scaling linearly with |
35 |
>> the CPU frequency. It was scaling *better* than that, because raising |
36 |
>> the FSB also made the mainboard itself perform better and with lower |
37 |
>> latencies. |
38 |
>> |
39 |
> and here we are - the point where the suspension of disbelief ends. |
40 |
> |
41 |
> All you may have gained you threw away with the slower ram - and you are |
42 |
> trying to tell us that your rig was faster? |
43 |
|
44 |
Yes. It made the difference in all games. I'm talking 40 vs 60FPS |
45 |
here. It was huge. |
46 |
|
47 |
The RAM wasn't much slower. Stock was 800 and I was running it at 756. |
48 |
|
49 |
|
50 |
> You do know that with today's CPUs the CPU is not the bottleneck - the |
51 |
> slow as molasses, no speed bump for 10 years ram is. |
52 |
> |
53 |
> (just look at the internal clock rate of dram chips - and you realize |
54 |
> that ddr1-3 are pretty much the same crap). |
55 |
|
56 |
The slightly slower RAM had no effect. As I said, the performance gain |
57 |
was huge. If the RAM ends up heavily underclocked to the FSB change, |
58 |
you just pick another ratio for it that brings it closer to its stock |
59 |
frequency, or slightly above it. Again, a good motherboard that has |
60 |
plenty of ratios to choose from helps immensely. |
61 |
|
62 |
Of course today this isn't important anymore. On my i5 CPU I can change |
63 |
the CPU multiplier. Not that I do; performance is plenty right now |
64 |
without OCing. I intend to overclock it in the future, just like I did |
65 |
with the C2D; if new games get more demanding, I'll do it then. |