I always buy the $100 processor and am always stuck unable to max my OC

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

PPB

Golden Member
Jul 5, 2013
1,118
168
106
if I had to make an updated OC/yield list on the 8xxx current lineup, It would look like this:

8370E>8370>8320E>8350=8310>8320. Discarding 9xxx series because they are high leakage parts, they do OC good but your cooling config wont handle it (and prolly your mobo wont either).

8370E being the absolutely best overclocker, 8320 being the worst. On that chain you got 2nd worst OCer SKU, having had a 8320 myself, I would say you can probably push your chip a little bit further into 4.5 territory. Not more than that. And if you do that, you are borderline close to your 4.7 target, the difference will be minuscule at best.

TL;DR: Push the chip a little harder into 4.5 with 4 modules activated. Add a backside fan to VRM/socket area, this helps a ton. Like someone really knowledgable said once, on FX 1/5 cooling needs to go to the VRM area, the other 4/5 to the CPU itself. Good VRM and backside cooling will help you a ton on your goal :)
 

monkeydelmagico

Diamond Member
Nov 16, 2011
3,961
145
106
TL;DR: Push the chip a little harder into 4.5 with 4 modules activated. Add a backside fan to VRM/socket area, this helps a ton. Like someone really knowledgable said once, on FX 1/5 cooling needs to go to the VRM area, the other 4/5 to the CPU itself. Good VRM and backside cooling will help you a ton on your goal :)

Really good point for OP to consider: His board is 4+1 VRM. And its a Gigabyte. Not exactly stellar track record for high voltages.

There is nothing about this mobo/cpu combo that is great for overclocking. He's lucky to get what he's got.
 

PPB

Golden Member
Jul 5, 2013
1,118
168
106
Oh didnt see his board. Yeah 4+1 is a stretch for OC. Dont overdo it with your board. And a higher binned chip wont save that fact.
 

DrMrLordX

Lifer
Apr 27, 2000
23,217
13,300
136
Uh? I don't think the ud4p has a 4+1 configuration. The ud3p sure doesn't.

edit: is there a ud4p? Google can't find one. It looks like a typo to me, and that he meant ud3p, which is an 8+2 board (technically, 4+1 with a doubler).
 
Last edited:
Dec 30, 2004
12,553
2
76
Really good point for OP to consider: His board is 4+1 VRM. And its a Gigabyte. Not exactly stellar track record for high voltages.

There is nothing about this mobo/cpu combo that is great for overclocking. He's lucky to get what he's got.

Oh didnt see his board. Yeah 4+1 is a stretch for OC. Dont overdo it with your board. And a higher binned chip wont save that fact.

Uh? I don't think the ud4p has a 4+1 configuration. The ud3p sure doesn't.

edit: is there a ud4p? Google can't find one. It looks like a typo to me, and that he meant ud3p, which is an 8+2 board (technically, 4+1 with a doubler).

yeah I have the 8+2 and a temperature software controlled fan mounted over the VRMs-- haven't had any VRM temp throttling since the fan upgrade. I am correcting my sig
 
Dec 30, 2004
12,553
2
76
Exactly. I think it's a down-binned 8320/8350, just like the 8300. Not a down-binned 8320e.

oh, ok. interesting. I just thought that basically all 8350s could hit 5ghz?

overclocking with a higher bus rate / lower multiplier does seem to make a difference. If I can get the temps managed I'm pretty sure I'll be able to hit 4.6ghz, and definitely 4.5. Even 4.5 I can notice being faster at day to day stuff...it's mainly Firefox that I notice it in.
 
Last edited:
Dec 30, 2004
12,553
2
76
alright I think I've got it stable at 4.6ghz with a bit of breathing room, just noisy fan. this definitely feels faster and feels good (fast) enough. I can also choose to either put up with the noise or upgrade the cooler if I want lower temps. I'll probably do this and see if I can't hit 4.7ghz. If I hit 4.7ghz, I'll probably realize I should just move to something Intel-based. Oh well, I enjoy the journey...

conclusions: overclocking with bus vs. multiplier definitely made a difference in hitting 4.6ghz, and I think giving the CPU PLL 2.6v instead of 2.5 also makes an ever so slight difference in stability-- it won't exactly help you hit something 100mhz higher, but if you're already there it provides a bit more stability without the heat increase of a higher VDD step.

I also just bought a kill-a-watt to record the consumption increase lol....
 
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
23,217
13,300
136
yeah I have the 8+2 and a temperature software controlled fan mounted over the VRMs-- haven't had any VRM temp throttling since the fan upgrade. I am correcting my sig

Okay, so it is the ud3p. Yeah you're covered for anything that isn't a 9590/9370. Theoretically I think the board could handle one of those, but the BIOS support isn't there.

oh, ok. interesting. I just thought that basically all 8350s could hit 5ghz?

Not really, no. 8350s are all over the map. You might get a golden chip, you might not.

The "chips to get" are 8370, 8370e, 8320e, and (for those that want them) wk29 2014 9590s (and later). There were a lot of people who thought the 8310 would do better than the 8320e since it had the same TDP with higher base clocks, but limited sampling suggests that the 8310 has inferior voltage scaling past 4-4.3 GHz.

8370e,8370, 8320e and 8310 all should be vishera-k

I haven't seen the term "vishera-k" used but I think I know what you're getting at. I have my doubts about the 8310. I could be wrong though.

I know that, I ... think I was imprecisely asking about what details of the e-process chips lead to them physically being better

The short answer to that inquiriy is that I don't exactly know. The long answer is full of assumptions.

Basically, you've got two groups of Vishera chips now (at least): pre 8370 and post 8370. When the 8370 hit, we also saw:

8370e
8320e
8310 (new version; OEM chips with this name but higher TDP have showed up before)
8300 (new version; see 8310)
Some 9590s that are better than older 9590s but have the same stock settings

It really isn't clear whether or not AMD kept cranking out 8320s, 8350s, 9370s, 63xx, and 43xx chips after they started turning out the above processors. My instincts are telling me that they stopped making those, or cut down on the production runs.

What differentiates the "new" Vishera chips is that GF seems to have tweaked the 32nm process for lower power consumption overall, and in some cases, higher clockspeed potential. It is not clear if all these "new" chips are coming off the same wafers or if they have different performance tweaks based on the desired end product. I lean towards the latter, though I have no proof.

If I am correct, then AMD gets the "better" 9590s and 8370s off wafers from production runs that do not emphasize low-leakage characteristics, and they get the 8320Es and 8370Es from production runs that do emphasize low-leakage characteristics. If I am not correct, then all four chips come from the same production runs.

It is also my assumption that the 8310 and 8300 do not come from wafers that benefit from GF's recent improvements to the 32nm process, and that they are excess inventory of 8320s or 8350s that have been rebadged and sold as OEM chips with lower clockspeeds and lower TDPs.

The net effect of these "new" chips is basically this:

The 8300 and 8310 seem to fare about as well as 8320s have clasically fared. You may get a lucky chip, but you may also crap out in the 4.5-4.7 GHz range.

The 8320e and 8370e overall outperform the average 8350 in terms of maximum overclock on air/water and in terms of voltage at a given clockspeed. 8370e chips have validated on water @ 5.7 GHz, which is pretty sick. There are people out there running 5 GHz 8320es with ~1.44-1.46v vcore. It still takes a monster board to pull off this trick, and temps on these chips are still difficult to control sometimes, but overall power draw seems to be a lot lower than your typical older 9370/9590 and a bit lower than a lucky 8350 that can do the same thing.

The 8370 is argued by some to be the better chip (compared to the 8730e) past 5 GHz no matter what the cooling, and it is almost undisputed to be better for LN2 runs.

What the "new" 9590 brings to the table is mostly unclear, though it seems these things just do the same clocks as the old ones with less voltage and less power draw. They are still binned as high-leakage parts, so they will still suck up more juice than other FX chips at the same vcore and clockspeed. Whether the 8370e, 8370, or "new" 9590 makes for the ideal clockspeed champion of Visheras is a matter for some debate. Given Vishera's age, it is unlikely that many people really care about that anymore.

tl;dr the 32nm process tweaks have breathed a little bit of new life into AMD's stangant FX lineup. My guess is the 8310 and 8300 simply don't benefit from that, though again, I could be wrong. It could really be that the 8300 and 8310 are just downbins and that all the "new" Visheras are coming from the same wafers, but given the considerable differences in leakage characteristics between, say the 8320e and the new gen 9590s, I have my doubts about there being a common source for all these processors.
 
Last edited:
Dec 30, 2004
12,553
2
76
I think given the insane difficulty I had reaching even 4.6 that that is a very likely possibility.

It would be nice to hit 5ghz but honestly given the heat I get now at 4.6 and those same voltages... I wouldn't hold out hope.

I definitely remember wishing I had just held on to my 720BE x3 that unlocked to x4 and overclocked to 3.5ghz. If I had I could have conceivably just used it in new mobo with the 32GB RAM (RAM the main reason I upgraded). In other words being at 4.6ghz now and only possibly maybe being able to get a little bit higher if I both bought another cooler and an 8370e...wouldn't really make much sense. Only way it would is if my current chip is just leaking power all over the place (which given temps is a distinct possibility in my opinion).

Also, I feel like I'm compensating for poor user experience and usability design in Chrome and poor optimization in Windows by just having a faster PC.
 
Last edited:

MongGrel

Lifer
Dec 3, 2013
38,466
3,067
121
I run my i5 at 4.6ghz using this cooler, and it's dead silent in my SFF computer case:

noctua_nh_l9i_2.jpg

Nice little Noctua, I've always been partial to those fans myself and never have used one of those.

ajJnwVI.jpg


The old X5680 needs more to get to 4.6

;)
 
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
23,217
13,300
136
Probably not until we start seeing boards in test/preview systems. Even then, it'll take a keen eye and the right kind of questions to figure out a detail like that. Usually that sort of question is not the first thing asked when a company like AMD demos a brand new microarchitecture on a brand new platform.
 
Dec 30, 2004
12,553
2
76
Probably not until we start seeing boards in test/preview systems. Even then, it'll take a keen eye and the right kind of questions to figure out a detail like that. Usually that sort of question is not the first thing asked when a company like AMD demos a brand new microarchitecture on a brand new platform.
lol YEAH YEAH YEAH WE DON'T CARE ABOUT 40% IPC WHAT IS THE MOUNTING BRACKET SIZE FOR THE HEATSINK IS IT SAME AS FM2 THX
 
Dec 30, 2004
12,553
2
76
Well, there's this:

http://forums.anandtech.com/showthread.php?t=2431539

(check post #4)

and . . .

http://www.hardwaresecrets.com/evga-superclock-cpu-cooler-review/6/

Bonzai usually knows what he's about, especially when he actually owns and uses the product. I would be more skeptical if he hadn't gone out and tried it for himself.

edit:

wrt lapping and the Ultra 120/TRUE, I was under the impression that this was important for various incarnations of the Ultra 120 due to uneven mating surfaces. The grit of the lap job was less important than just getting the darn thing to be flat. I lapped my D14 to 800 grit and it helped a little. Only reason why I lapped my 7700k was that I had the lid off anyway, so why not?

Anyway mirror finish isn't going to do you much good over an 800 grit lap. It might actually be a hair worse.

of course, the day after I buy this Anandtech releases a new cooler review. go figure

that's a pretty bold claim, to be able to beat the NH-D15. I hope I didn't just throw my money away
 

CHADBOGA

Platinum Member
Mar 31, 2009
2,135
833
136
of course, the day after I buy this Anandtech releases a new cooler review. go figure

that's a pretty bold claim, to be able to beat the NH-D15. I hope I didn't just throw my money away
It only edges out the NH-D15, if you can live with a ridiculous amount of noise from the higher speed fans.

Otherwise the NH-D15 is still the best heatsink from that review.
 

DrMrLordX

Lifer
Apr 27, 2000
23,217
13,300
136
Anytime you have an HSF that ships with fast fans, you have an HSF that may be trying to edge out the competition at the expense of noise from said fans. You could just as easily add faster fans to some other HSF (such as the D15) to close the performance gap.
 
Dec 30, 2004
12,553
2
76
Anytime you have an HSF that ships with fast fans, you have an HSF that may be trying to edge out the competition at the expense of noise from said fans. You could just as easily add faster fans to some other HSF (such as the D15) to close the performance gap.
I made a bad purchase. what should I do?
 
Dec 30, 2004
12,553
2
76
If you bought the EVGA ACX, then you didn't make a bad purchase. You made a good purchase. Just don't run the fans wide open.

oh, alright. Glad to hear BonzaiDuck isn't the only one who thinks it's good. I'll try to rig up some semi-scientific before/after test between my Ultra-120 with AS5 and the EVGA ACX. I'm sure people would like a double test of the AS5 and the NT-H1 but. Actually, I will make a post about that