X79 and/or rampage IV OC'ing thread for those of us still left....

[dalekdukesboy]

I used to have long telephone conversations with The RAM Guy at Corsair. He told me several years ago the X79 memory controllers were the weak spot in the system. His advice when running large amounts of RAM like I usually do is to increase the voltage to 1.65 volts for DDR3. He said that eases the strain on the controller. So far, so good on that advice.

I began experimenting with case cooling 20+ years ago (and carved up countless cases with the Dremel and plasma cutter) and am a firm believer in the better airflow is the longer things last, especially when overclocking. My favorite cases are still the Corsair Carbide 500R series because of their great design and lack of need to do any cutting for proper fan placement.

Well I admit I've done a fair amount of hacking and crazy setups, this one definitely included....I'm liking the results that I'm getting but wish I could get a bit cooler on the processor though what I use for stability testing is Linx/burntest which literally runs the CPU hotter than anything else you would ever run including stability tests Prime 95 etc....so for starters do you (or anyone else who wants to chime in) think Linx is even best thing I could use for stability testing or is it overkill/unnecessary?

Also...I have 2 rads in my setup one on top and one in front, one is a 280 mm I believe and the one on top is a 420 mm or something like that, it literally just fits into the top of the case. I have only a Push setup on both I have 3 fans on one and 2 on the other....fans are all the 140mm Silverstone fans which are rated for 175 cfm I believe. They were quietest fans with highest CFM rating I could find. Do you think I'd gain much if I could get some "pull" fans on one or both rads? I notice you have push/pull and I've played around with it before with varying results, just not sure with watercooling if more fans is somewhat useless after a certain point or if more is better?

 

[bobbybluz]

Those 140mm Silverstones are the FHP141's. They do 171cfm at full speed. I have two of those in the side panel of both my Corsair 500R's. Here's a tip: The two speed selector switches on those wonk out after a while. The contacts corrode and they'll default to low speed only. I have 6 of them in use at the moment and had the switch problem on all of them at various times. My solution was to solder the high speed leads on the switches together so the actual switch is bypassed. Make sure to only solder the correct ones.

My stress tests are video rendering and transcoding. I found Gelid GC Extreme to be the best solution for thermal compound for my needs. It's cheaper and far easier to apply than the liquid metal stuff plus isn't electrically conductive and can easily be wiped off the CPU heatspreader. If you're running those FHP141's at full speed I doubt going push/pull would be even worth considering. I found that mounting the radiators outside the case is very effective at reducing temperatures and it's free. That's why I like the 500R because no cutting is involved for a top mount, just pop the panel out and you're good to go. Remember that there are diminishing returns in cooling and no matter how much radiator you have you'll always be limited by the thermal transfer capability of the pump and plate itself. A decent 280mm radiator with two FHP141's in push or pull should be more than adequate for a 1680 V2 with 1.42 or less core volts at 4.7GHz or less clock speed provided you're pulling enough air into the case. From my experience and perspective a 420mm radiator is overkill for this application all factors considered.

 

[dalekdukesboy]

Those 140mm Silverstones are the FHP141's. They do 171cfm at full speed. I have two of those in the side panel of both my Corsair 500R's. Here's a tip: The two speed selector switches on those wonk out after a while. The contacts corrode and they'll default to low speed only. I have 6 of them in use at the moment and had the switch problem on all of them at various times. My solution was to solder the high speed leads on the switches together so the actual switch is bypassed. Make sure to only solder the correct ones.

My stress tests are video rendering and transcoding. I found Gelid GC Extreme to be the best solution for thermal compound for my needs. It's cheaper and far easier to apply than the liquid metal stuff plus isn't electrically conductive and can easily be wiped off the CPU heatspreader. If you're running those FHP141's at full speed I doubt going push/pull would be even worth considering. I found that mounting the radiators outside the case is very effective at reducing temperatures and it's free. That's why I like the 500R because no cutting is involved for a top mount, just pop the panel out and you're good to go. Remember that there are diminishing returns in cooling and no matter how much radiator you have you'll always be limited by the thermal transfer capability of the pump and plate itself. A decent 280mm radiator with two FHP141's in push or pull should be more than adequate for a 1680 V2 with 1.42 or less core volts at 4.7GHz or less clock speed provided you're pulling enough air into the case. From my experience and perspective a 420mm radiator is overkill for this application all factors considered.

Thanks, I've had fans for quite a while and they've all worked at full speed I think they are couple years old...however recently I hadn't checked them but oddly enough with your tip I checked them all and 3 hum along great and 2 of them don't react with switch and whatever that speed is they are running at, is crawling I can put my finger into fan and it stops with no effort whatsoever, I obviously touched center of fan first to see how "fast" it was spinning and it was so slow I knew you couldn't get cut or have the fan blade damaged from hitting something if you tried! I'm not even sure they do the low speed when switch goes. Although I just looked it up they spin at 500 rpm's and do 42 cfm lol, that is almost nothing for a fan that size. I'm pretty sure it happened just recently because I had switched them all not too long ago and they worked, and just this week I swore my computer got "quieter" and with two of those fans going that slow well that's the obvious reason why!

So thanks for that advice! I know it's overkill but it is also a big case and believe it or not I did at one point test the difference between 1 rad and 2 and there was a definitive difference. 1 rad is an all in one from OC cooling that is the 280 mm one and I just so happened to have a 420 mm rad hanging around which was also an AIO cooler from same company but it came with a sadly weak pump so I just removed pump and the tubing all fit together nice and got it all in one case. I have read a lot about WC'ing and realize there are diminishing returns with rad size/fans and the plate performance plus pump speed/pressure would be even more important.

Anyway now I have results with those 2 fans putting along at slow speed I'm going to try your fix on them or get two other fans in their place either way will be interesting to see if it makes much difference. Yeah I had MX-4 for the compound so I used it, it's pretty good stuff and AC5 is somewhat dated now so I thought it was a good performing paste, though not sure how it compares with Gelid's paste performance wise though.

 

[Kanan]

So, I'm back.

Gonna update you guys:

Had the Core i7 3960X, as some people know. Had some issues with it in BF5, GPU was maxing out at only 50% util in DX11 and 75% in DX12, which is managable, but not exactly good. So this led me to plan a Ryzen 4000 upgrade. Then the motherboard decided to go degraded on me, resulting in full freezes, crashes. Had to start the PC 2 times before I did anything, as a work around to prevent a full freeze. That is all at non-spectacular 4500 MHz. The CPU was kinda golden, could do 5 GHz easily, 5.1 possible, and maybe even more, couldn't test without a water cooler.

Anyways, because of that I pulled the gun early and went with Ryzen 3700X instead. Results: GPU bottlenecks gone, BF5, GTA 5 even more so, this game is kinda ancient when it comes to CPU usage, it needs high IPC. All problems gone, no "real" OC needed, as Ryzen is quite intelligent, and PBO will do it automatically - technically a overclock too, but not a traditional one. CPU maxes out at 150 W in extreme Prime 95. Without PBO it stays at 95W max, according to its TDP. Normal usage with PBO, 110W max. TimeSpy, 115W max. Is pretty good, I can only recommend it. New platform is great as well, RGB looks good, not an important factor, but the board is great, has better power delivery than my X79 board had, which is a surprise. Newest USB, NVME, great things. Onboard sound was quite good as well, but I wanted to stay with Soundblaster, so I bought a new one, because this one does not support my PCI based X-Fi. All right. Too much talk, I'll provide a link too:

Result

www.3dmark.com

IPC uplift is quite heavy, 2 cores alone can not do this. Yeah, best thing ever, got rid of overclocking headaches, this CPU will never be unstable because it manages itself, even if overclocked.

So long...

 

[karakarga]

@Peter Lindgren

The best possible way of increasing x79 platform performance is using Xeon E5-2696 v2 CPU. Overclocking Xeon E5-1680 v2 consumes too much power, maybe around 200 Watts. Xeon E5-2696 v2 consumes only 120 Watts. At aliexpress, Xeon E5-2696 v2 prices dropped around $130. I have double from those on my Z9PE-D8 WS. They have 9250~9500 points on CPU-Z. It is currently pricey and very difficult to find an Asus Z9PE-D8 WS, but Jingsha X79Dual S8 is easy to obtain. This one is around $155.

An alternative way of increasing system performance is, using Xeon E5-2678 v3 CPU on Huananzhi X99-T8 mainboard with DDR3 memory. (There are X79-TF and X79-F8 as well.) Together, those will cost around $200. There is also a double CPU option by using 2678 v3 CPU's again with DDR3's on Huananzhi X99-T8D mainboard. That mainboard is a bit pricey right now $210, but gives roughly the same performance of Z9PE-D8 WS with Xeon 2696 v2 CPU's.

Asus Z9PE-D8 WS ($600) + 2 x E5-2696 v2 $ (2 x 130) = $860
Jingsha X79Dual S8 ($155) + 2 x E5-2696 v2 $ (2 x 130) = $415 *
Huananzhi X99-T8D ($210) + 2 x 2678 v3 (2 x $97.5) = $405 *
Huananzhi X99-T8 ($105) + 1 x 2678 v3 ($97.5) = $202.5 **

* lower than half of the price of the above configuration with same performance. (I assume you already have 2 to 8 pieces of DDR3 Rams.)
** lower than quad the price of the above configuration with half the performance. (I assume you already have 1 to 8 pieces of DDR3 Rams.)

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Update: Huananzhi X99-T8D prices dropped $135.15 including shipping. So,

Asus Z9PE-D8 WS ($600) + 2 x E5-2696 v2 $ (2 x 130) = $860
Jingsha X79Dual S8 ($155) + 2 x E5-2696 v2 $ (2 x 130) = $415 *
Huananzhi X99-T8D ($135) + 2 x 2678 v3 (2 x $97.5) = $330*
Huananzhi X99-T8 ($105) + 1 x 2678 v3 ($97.5) = $202.5 **

 

[Zyll Goliat]

So, I'm back.

Gonna update you guys:

Had the Core i7 3960X, as some people know. Had some issues with it in BF5, GPU was maxing out at only 50% util in DX11 and 75% in DX12, which is managable, but not exactly good. So this led me to plan a Ryzen 4000 upgrade. Then the motherboard decided to go degraded on me, resulting in full freezes, crashes. Had to start the PC 2 times before I did anything, as a work around to prevent a full freeze. That is all at non-spectacular 4500 MHz. The CPU was kinda golden, could do 5 GHz easily, 5.1 possible, and maybe even more, couldn't test without a water cooler.

Anyways, because of that I pulled the gun early and went with Ryzen 3700X instead. Results: GPU bottlenecks gone, BF5, GTA 5 even more so, this game is kinda ancient when it comes to CPU usage, it needs high IPC. All problems gone, no "real" OC needed, as Ryzen is quite intelligent, and PBO will do it automatically - technically a overclock too, but not a traditional one. CPU maxes out at 150 W in extreme Prime 95. Without PBO it stays at 95W max, according to its TDP. Normal usage with PBO, 110W max. TimeSpy, 115W max. Is pretty good, I can only recommend it. New platform is great as well, RGB looks good, not an important factor, but the board is great, has better power delivery than my X79 board had, which is a surprise. Newest USB, NVME, great things. Onboard sound was quite good as well, but I wanted to stay with Soundblaster, so I bought a new one, because this one does not support my PCI based X-Fi. All right. Too much talk, I'll provide a link too:

Result

www.3dmark.com

IPC uplift is quite heavy, 2 cores alone can not do this. Yeah, best thing ever, got rid of overclocking headaches, this CPU will never be unstable because it manages itself, even if overclocked.

So long...

Well yeah sure there is no doubt that Ryzen 3000 series are great CPU's but IF you really want to compare Ryzen with the X79 then you can maybe do that with the first-gen Ryzen and I did that and choose to go with the Xeon&X79 instead....it cost me less + I get way better quality mobo and much more memory and the performance is more or less the same.....Again really depend what someone is looking for....some people are just nostalgic some people still prefer quad channel ddr3 over dual channel ddr4,some people love new hardware...others prefer older...some will always go for best price for performance ....and so on....

 

[Kanan]

Well yeah sure there is no doubt that Ryzen 3000 series are great CPU's but IF you really want to compare Ryzen with the X79 then you can maybe do that with the first-gen Ryzen and I did that and choose to go with the Xeon&X79 instead....it cost me less + I get way better quality mobo and much more memory and the performance is more or less the same.....Again really depend what someone is looking for....some people are just nostalgic some people still prefer quad channel ddr3 over dual channel ddr4,some people love new hardware...others prefer older...some will always go for best price for performance ....and so on....

No dude, this is only my personal experience, it is not about making a point or something like that. Also yes, 1st Gen Ryzen is not a upgrade to X79 based CPUs, not even 2nd gen is, unless you don't overclock. I pretty much ignored those two. PCs imo are just about doing things, they are tools. IMO people should not get too attached with these things. I'm not interested in a fanboy debate, I posted my experience and I'm more or less finished here. Edit: It is also about a discussion, we had much earlier, you were not part of that.

 

[Zyll Goliat]

No dude, this is only my personal experience, it is not about making a point or something like that. Also yes, 1st Gen Ryzen is not a upgrade to X79 based CPUs, not even 2nd gen is, unless you don't overclock. I pretty much ignored those two. PCs imo are just about doing things, they are tools. IMO people should not get too attached with these things. I'm not interested in a fanboy debate, I posted my experience and I'm more or less finished here.

Cool...don't get me wrong I am not the fanboy either I will always go for price/performance instead really do not care if is it Intel,AMD,Nvidia.....

 

[dalekdukesboy]

This is the little chip that could....

 

[phill]

Very impressive there @dalekdukesboy !! I'd love a 1680 V2, I just can't stomach the price to buy one!!

 

[dalekdukesboy]

Only 1 gig of memory but still....10 runs at 4.8 ghz....wow.

Very impressive there @dalekdukesboy !! I'd love a 1680 V2, I just can't stomach the price to buy one!!

Yes it is, this thing is a beast of a chip. Even with 20 bucks extra to get it to me quick it only cost me $233 US dollars which really isn't all that much for what you're getting.

Not bad voltage either. I just gamed for half hour or so with it just to see and so far so good, I'm amazed.

Well it's definitely benchmark stable and it can do 5 runs linx with 2 gig memory out of 16 installed so not bad, not bad at all.

 

[Kanan]

Here are more comparisons, Old VS New - this is not a pissing contest, IDGAF about those, this is scientific / practical data for those who want to switch now or some day and want to see what happens when they do. Context: I switched from a i7 3960X EE to a Ryzen 3700X, because the mainboard gave up on me. RAM is very comparable, actually the old one had a tad more bandwidth, this is kinda relevant too. This is 1866 Quad Channel vs 3600 Dual Channel. What is also interesting, this is chiplet based vs monolithic, good old ringbus vs external IODIE.

3DM11 (old vs new):

Result

www.3dmark.com

3DM Firestrike:

Result

www.3dmark.com

3DM Vantage Extreme (DX10):

Result

www.3dmark.com

(for the sake of completeness again)
3DM Timespy: (3960X @ 4500 MHz)

Result

www.3dmark.com

Keep in mind that in those older ones, the 3960X is clocked on 4800 MHz, something I didn't use for day to day. For TimeSpy it was more realistically clocked at 4500 MHz, so TimeSpy is the best reflection on "today" performance, aside from the clock, too. DX12 is the future, I think.

PS. If nobody cares about this data, and is just offended, I can delete this, I'm not here to "offend" anyone. This is a excellent comparison, because users of X79 can not forever stay on a old platform that is slowly but surely dying. I used X79 for 7 years.

 

[k_9virus]

dang ill do some benchies too with my TR1900x vs my 3960x lol

@Kanan

 

[Kanan]

dang ill do some benchies too with my TR1900x vs my 3960x lol

@Kanan

yeah, maybe even more relevant, because it is HEDT vs HEDT.

 

[Zyll Goliat]

Here are more comparisons, Old VS New - this is not a pissing contest, IDGAF about those, this is scientific / practical data for those who want to switch now or some day and want to see what happens when they do. Context: I switched from a i7 3960X EE to a Ryzen 3700X, because the mainboard gave up on me. RAM is very comparable, actually the old one had a tad more bandwidth, this is kinda relevant too. This is 1866 Quad Channel vs 3600 Dual Channel. What is also interesting, this is chiplet based vs monolithic, good old ringbus vs external IODIE.

3DM11 (old vs new):

Result

www.3dmark.com

3DM Firestrike:

Result

www.3dmark.com

3DM Vantage Extreme (DX10):

Result

www.3dmark.com

(for the sake of completeness again)
3DM Timespy: (3960X @ 4500 MHz)

Result

www.3dmark.com

Keep in mind that in those older ones, the 3960X is clocked on 4800 MHz, something I didn't use for day to day. For TimeSpy it was more realistically clocked at 4500 MHz, so TimeSpy is the best reflection on "today" performance, aside from the clock, too. DX12 is the future, I think.

PS. If nobody cares about this data, and is just offended, I can delete this, I'm not here to "offend" anyone. This is a excellent comparison, because users of X79 can not forever stay on a old platform that is slowly but surely dying. I used X79 for 7 years.

Well yeah...but then again this is also X79 for example

and definitely more relevant as it is 8c/16t Vs 8c/16t.....CPU score is almost identical and that particular 1680 V2 was OC on 4,751mhz

 

[Kanan]

Well yeah...but then again this is also X79 for example
View attachment 156594
and definitely more relevant as it is 8c/16t Vs 8c/16t

Yep, this is a 350€ CPU. And it is highly overclocked and inefficient compared to my 3700X, which only uses about 110 W achieving these scores. In other words: Ryzen 3000 is still highly better.

Wait, did you really think I did not do my homework, or consider this CPU for myself? Tip: read up on old posts in this board.

Money wise comparisons: I would add 50€ and get the 3900X and destroy this result you posted. And then enjoy my low power consumption on a platform that is actually modern and has NVME.

 

[Zyll Goliat]

Yep, this is a 350€ CPU. And it is highly overclocked and inefficient compared to my 3700X, which only uses about 110 W achieving this scores. In other words: Ryzen 3000 is still highly better.

Wait, did you really think I did not do my homework, or consider this CPU for myself? Tip: read up on old posts in this board.

Uhhh....srry if you get offended or something ....1680 V2 is not anymore that expensive anyway....again sure 3700x is better no doubt but that old "mule"is still running a race.....GL with your new red shirt....

 

[Kanan]

Uhhh....srry if you get offended or something ....1680 V2 is not anymore that expensive anyway....again sure 3700x is better no doubt but that old "mule"is still running a race.....GL with your new red shirt....

I don't think I'm offended, because you have the urge to force discussions like these. I'm just answering. And I've just checked ebay, the cheapest offer is 337€ in France. This is a expensive USED CPU for a dead platform.

 

[k_9virus]

PS. If nobody cares about this data, and is just offended, I can delete this, I'm not here to "offend" anyone. This is a excellent comparison, because users of X79 can not forever stay on a old platform that is slowly but surely dying. I used X79 for 7 years.

i care btw based on scores you posted since i still have this platform.

 

[Kanan]

i care btw based on scores you posted since i still have this platform.

You have a TR platform and X79?

 

[Zyll Goliat]

I don't think I'm offended, because you have the urge to force discussions like these. I'm just answering. And I've just checked ebay, the cheapest offer is 337€ in France. This is a expensive USED CPU for a dead platform.

I am not forcing anything this is clearly forum about hardware and thread is about X79 platform....if anything discussion about this subject is a MUST....again srry if you get offended or something....
also check this PRICE of 1680 V2 and there is more similar prices that are bellow or close to the 200$....

 

[Kanan]

I am not forcing anything this is clearly forum about hardware and thread is about X79 platform....if anything discussion about this subject is a MUST....again srry if you get offended or something....
also check this PRICE of 1680 V2 and there is more&more like that......

From the get go you reacted in a defensive way and forced unnecessary discussions, told me, my comparison is bad, etc. so yes you did. I have already checked the price, maybe read what I write for a change.

 

[k_9virus]

You have a TR platform and X79?

yup first gen TR 1900x and i7 3960x ( i need to put the I7 because theres a TR 3960x lol)

 

[Kanan]

@Zyll Goliath I'm not buying something from US, ever heard of customs and shipping prices? It has to be absurdly cheap to be worth the hazzle, which it is not.

Why aren't you buying it yourself, if it's so "cheap"? "Xeon 2650 V2 (OC-Base Clock 3,4Ghz-Turbo 3,8Ghz) " great CPU you have, even a 50€ CPU is better than this for gaming.

yup first gen TR 1900x and i7 3960x ( i need to put the I7 because theres a TR 3960x lol)

Looking forward to comparisons then! edit: yeah the naming is actually pretty funny, new 3960X is a beast too

 

[k_9virus]

@Kanan

problem is i need to use 1 GPU for comparison my vega64 is in watercooling loop. once im due for maintenance ill do some benchies for now here's some screenshots