1366 X58 Xeon 5650

KungFuSpaghetti · 11 Dec 2017 at 06:10

As you say buddy, the only ones it seems can be messed with are the e5 16xx series. An e5 1680 v2 is about the same price used as a brand new i7 8700k, my system is purely a gaming machine, so looks like I'll be holding on to the X series a while. If prices bottom out on the newer xeon it might be an option. I do love the bargain basement performance!

Saffy · 11 Dec 2017 at 06:58

I'm with you!

Did you manage to get things stable with the new settings?

KungFuSpaghetti · 11 Dec 2017 at 07:25

Only if I set my memory to auto... Otherwise it won't post. Think the ram I have is causingthe stability issues.

Saffy · 11 Dec 2017 at 07:54

KungFuSpaghetti · 11 Dec 2017 at 08:28

I can get to 4.8 stable, but have to put memory on auto. Doing that it's running in the 1200's though. Cinebench 11.5 I'm getting 11.87 and 1096 on R15. Temps are 74 degrees max on core 5... Not sure what to do... Lol

DragonQ · 11 Dec 2017 at 10:48

Saffy said:
I'm in the same situation. I have both 1.5v and 1.65v DIMMs. What speed were you running yours at?

I have mine at just over my slowest DIMM at around 1650MHz.

It would've been around 1600 MHz, but I took out the old sticks to leave me with just 16 GiB at 1866 MHz using 1.5 V.

Saffy · 11 Dec 2017 at 10:55

KungFuSpaghetti said:
I can get to 4.8 stable, but have to put memory on auto. Doing that it's running in the 1200's though. Cinebench 11.5 I'm getting 11.87 and 1096 on R15. Temps are 74 degrees max on core 5... Not sure what to do... Lol

That's really good! Is that on the D14 cooler?

I'd run it at a dead certain stable, regardless of RAM compromise to be honest. I don't think there is a massive RAM advantage to be had. Unless you buy a "new" kit altogether.

DragonQ said:
It would've been around 1600 MHz, but I took out the old sticks to leave me with just 16 GiB at 1866 MHz using 1.5 V.

That's around the same as mine then. I think I'll keep it around that and keep an eye on sticks exactly like the fastest I have.

KungFuSpaghetti · 11 Dec 2017 at 13:29

Yeah this is on the d14 with the define C fans in stock position, so one drawing air to the cooler and one behind venting it out. Temps never seem to be the problem, it's as if my board or chip hits 210+ BCLK then instantly becomes unstable. 4.7 or above and it posts fine, boots to Windows and can run Cinebench a couple of times okay, but play a game or run ibt for 20 mins and BSOD

Saffy · 11 Dec 2017 at 20:11

That's probably an indication of reaching its upper limit. Still, even if you end up stable at 4.5/6GHz, that's pretty good. I got a multi score of 1040 at 4.5GHz in Cinebench, which isn't too bad at all, but temps were high. I might buy the D15 as a replacement cooler because 1. it's a good investment 2. I kinda need the cooling for higher clockspeed.

I messaged Noctua to ask if they still offer the free socket 1366 kit and they said yes.

So if all goes well, I'll get a "powerhouse" cooler and run 4.5GHz.

KungFuSpaghetti · 11 Dec 2017 at 20:59

Yeah tried again to get 4.7 to 4.8 stable. I can get them stable ish with 1.5vcore with core 5 at 76 degrees, the rest at 70-71. Only thing so far I've noticed is i can no longer run cinebench, it just crashes. Games, ibt and prime 95, all fine... Wtf!?

D15 will be awesome, didn't even know they offered a bracket, that's why I went for the D14... Mind you it was cheaper lol.

Saffy · 11 Dec 2017 at 21:06

It's not stable.

How many passes of IBT have you ran and which setting? Are you using all the RAM available?

I've read somewhere that 25 passes on the maximum setting is a good indication of stability. And 50 passes being the ideal number if you don't want to do a 24/7 test.

I ran prime95 for a long while without an error. I then ran IBT and got an error on the 3rd pass. So I can conclude that prime95 isn't very good for stabilty testing. However, it is a good test for thermals with the small ftt test.

KungFuSpaghetti · 12 Dec 2017 at 08:31

Yeah it's deffo not stable, crashes out on most apps so have just put it back to 4.5 for the time being, but with memory all on auto. Ran on max settings prior to this and it does about 8-10 passes before BSOD

I think my RAM being 1600mhz but poor timings isn't helping matters, but hard to justify spending to replace it. I'm wondering if it's worth putting the memory timings in manually rather than leaving it all auto, also wondering if it will take different timing settings, or are these locked in to the specifics of the memory?

Saffy · 12 Dec 2017 at 08:40

Sorry to hear that.

It's worth trying all combinations to see if stability can improve. If its blue screening so soon then it won't be long until you find the right combo... hopefully.

I've ordered a new D15 cooler. It wasn't initially part of my budget since this was a budget orientated build but it's a good investment long term I suppose.

Ste · 12 Dec 2017 at 10:12

PCIe to 101 could help with some stability when messing with higher bclk ranges though it will break USB3

Mine was only ever 20 mins prime stable at 80c+ but I gamed on it fine for two years

Saffy · 12 Dec 2017 at 11:06

Ste said:
PCIe to 101 could help with some stability when messing with higher bclk ranges though it will break USB3

Mine was only ever 20 mins prime stable at 80c+ but I gamed on it fine for two years

I agree with the 20 mins stability, especially for gaming. Games hardly stress the CPU compared to any given stress test. But, 2 years is a long time.

AiiR · 12 Dec 2017 at 16:18

Born2BSlow said:
Well I finally upgraded from my Xeon 5670 @4.2 to an i5 8600k clocked to 4.6, all that money spent on all new kit... Guess what! my 3DMark physics score is 2.5% lower than my old Xeon system. I daren't bring it back downstairs and stick the 1080Ti in it, the missus would go ballistic lol. Ah well, this will last me 5 years easy, and I reckon the Xeon will still be ticking along nicely then as well.

When I went from a 5650 at 3.6ghz to a 6600k at 4.6ghz my Physics score on the 6600k was slightly lower but overall there was a pretty noticeable improvement in frame rates. I wouldn't pay too much attention to it.

KungFuSpaghetti · 13 Dec 2017 at 08:24

Well I've settled on my OC now at 4.5ghz stable, just going through reducing voltage now to get it as low as possible, which is currently 1.4v vcore. Going to leave it like this now, as I made a couple of changes to the BIOS OC settings and I'm getting 11.19 in Cinebench 11.5 and 1015 in R15. Happy with those and CPU-Z bench reports it running just under a Ryzen 1600 on multicore and significantly higher on single thread. Combining that with my 1070ti OC'd by 250/500 it's a quick system

Saffy · 13 Dec 2017 at 12:13

That's a great system you have there!

I need more voltage at 4.5GHz. However, in Cinebench R15, I get the same score but at 4.4GHz with RAM at the adjusted speed (after BCLK increase). 1040 is the score I get at 4.5GHz. Have you underclocked your RAM?

You can get a faster score with RAM running at the adjusted speed. But if its an issue at that speed then I suppose running it at where it is the best option. Still, it's a very capable spec.

KungFuSpaghetti · 13 Dec 2017 at 14:41

Thanks Saffy, for a budget build I'm pretty pleased with it and the GPU makes up for any shortfall in CPU/RAM performance. Then again, I don't think my CPU is a bottleneck, far from it in fact as I get better FPS score than my mates 1700 Ryzen with a Gigabyte Gaming 1080 (his is stock, mine is OC'd by 250/500). The RAM is certainly my bottleneck though, as it wont post if I manually set the speed so it's running around 1600mhz (which is what my RAM is!?!?). I think my timings are 9-9-9-24 for the memory, but have left them in auto for everything at the minute. Not sure if it's worth manually setting them?

Saffy · 13 Dec 2017 at 15:23

I've believed for a while that Ryzen's aren't (ideal) for gaming and your statement proves that once again.

I mean, they're getting left behind in the dust by, let's face it, a legendary but ancient Xeon. AMD have to do more, like putting all their chips in the gym to work on that single core performance.

I'm guessing that leaving your RAM on auto will make it work no matter the scenario. However, I don't know if manually adjusting the timings will have a similar effect. It's worth a try though.

Have you ran the CPU mark test? I'd be curious to know how your chip performs at 4.5GHz. My overall score at 4.5GHz was 10404, while the single threaded score was 1947. That's an improvement of 33% (multi) and 37% (single). A massive jump over stock speeds. I'm not sure how RAM affects the scores, but I know that it definitely does make a little difference in Cinebench R15.