|
|
Post by ShrimpBrime on Dec 7, 2020 18:20:11 GMT -5
Need some advice on voltage scaling with 4000+ ClawHammer.
Last I had it running, 1.620v at 2688mhz, But I think I was using a not so good memory divider.... Not sure, I have only 1 bench with the chip. Couldn't keep focused on it at the time.
Would it be feasible to run say 1.8v - 1.9v on sub ambient but above freezing on 130nm?
(The cpu is not installed yet. Had to clean it up, has a couple wavy pins. Maybe tonight I can swap chips after the repair.....)
|
|
|
|
Post by Mr.Scott on Dec 7, 2020 21:03:17 GMT -5
Claws don't clock.
Sub ambient above freezing you'll be lucky to see 2.8-2.9
For voltage, probably 1.7-1.8v
180 divider with good ram.
|
|
|
|
Post by ShrimpBrime on Dec 7, 2020 21:58:10 GMT -5
Claws don't clock. Sub ambient above freezing you'll be lucky to see 2.8-2.9 For voltage, probably 1.7-1.8v 180 divider with good ram. Cool thanks for the heads up. Might pop it in yet, looks like it'll seat the socket well now. I don't have but one sub, so thought to try this one out. Was wondering if you think a 10x or 9x multi would be the way to go? The 148 seemed to like 9x and over 1.7v. For certain benches though. ( I know totally different cores) |
|
|
|
Post by ShrimpBrime on Dec 7, 2020 22:01:28 GMT -5
Oh and the HT, do you think 8x would be better suited vs 16x say for max clocks or any benchmarks?
Note: I don't think I've ever tried 8x for HT bandwidth, so that's why I'm asking.
|
|
|
|
Post by Mr.Scott on Dec 8, 2020 19:24:03 GMT -5
Never tried. I always preferred 11x because it maxed or near maxed my ram using a relatively high FSB. Experiment. It ain't gonna hurt. GL and good clocking.  |
|
|
|
Post by ShrimpBrime on Dec 8, 2020 23:38:11 GMT -5
Never tried. I always preferred 11x because it maxed or near maxed my ram using a relatively high FSB. Experiment. It ain't gonna hurt. GL and good clocking. Right, I'll have to experiment with that. Did give it a go last night, did some minor tweaking stuff here and there. Went for clocks right from the box. Terrible results with low Bus and high multi in regards to running PiFast. Max clock, I need to validate yet. 2933mhz, not a stitch over or get a blue screen. "Not bad, but not too good either." seems average and right where you said it would be. Benching in the 2.8ghz area. Is fine, I'm having fun. See if I can't get a screen shot or something in here in a little bit. |
|
|
|
Post by ShrimpBrime on Dec 9, 2020 11:38:40 GMT -5
Is this a thing?
|
|
|
|
Post by Mr.Scott on Dec 9, 2020 16:31:12 GMT -5
It's like I made that vid.
Basic tweaks for Pifast on a Claw with BH-5 ram. Or any 939 with BH-5 for that matter.
I actually tweak a little harder than that.
|
|
|
|
Post by ShrimpBrime on Dec 9, 2020 17:37:08 GMT -5
It's like I made that vid. Basic tweaks for Pifast on a Claw with BH-5 ram. Or any 939 with BH-5 for that matter. I actually tweak a little harder than that. The 148 will do 2-2-2-0-9. This 4000+ with the same timings just boot cycles. Probably could run the 150 divider and tune the sub timings.... However. The board with the nodded bios, as you saw in the beginning was at defaults. Way strange timings. When set to auto all. 2-3-2-5-12 are the timings that belong there. I did reflash the board, but clear cmos sets 4-0-4-4 or w/e. The only sticks it will post with are my XMS. The Adata wont post at all at defaults. Gotta set profile and then hot swap the memory, can get the Adata to run. Havent tested them on this chip yet. |
|
