![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Intel Skylake Core i7-6700K IPC & Overclocking Review @ [H]
- Thread starter FrgMstr
- Start date
this seems like a good value Zaniix:
http://www.ncix.com/detail/g-skill-ripjaws-v-series-f4-3200c16d-16gvk-04-111442.htm
( specs here http://www.gskill.com/en/product/f4-3200c16d-16gvk )
It seems to me that the vast majority of kits are 4x4gb. 2x8gb seems to be an exception, let alone 2x16gb. sheesh.
I'm okay with 16gb of RAM for now. Would be nice to have more, later.
http://www.ncix.com/detail/g-skill-ripjaws-v-series-f4-3200c16d-16gvk-04-111442.htm
( specs here http://www.gskill.com/en/product/f4-3200c16d-16gvk )
It seems to me that the vast majority of kits are 4x4gb. 2x8gb seems to be an exception, let alone 2x16gb. sheesh.
I'm okay with 16gb of RAM for now. Would be nice to have more, later.
4RunnerTE
Limp Gawd
- Joined
- Aug 12, 2015
- Messages
- 222
Any difference between this "new" ripjaws V and the 4? Can you still use the 4 on any z170 chipset?
Yeah, the problem is that a DDR4 3000 2x8 kit will be $50 - $100 less and I have yet to see any data saying that is worth going 3200 over 3000
My decision will most likely be driven by what is available at the best prices when I pull the trigger. Since I cant buy the CPU yet I am waiting on MB and RAM unless I see some amazing deal I cant pass up..
I have already had to much time to think about the Mother Board choices. Now I am counting how many PWM fan headers one board has verses the next. I have officially gotten silly in my choices. I should probably just buy a decent Fan controller and then I can stop worry about it. Surprised me that the MSI Z170A GAMING M7 only has 4 fan headers. I would think it is standard to expect 2 fans for the Heatsink or AIO, 1 exhaust and 2 intake. Anyway enough of my crazy OCD.
Back to searching for MB reviews
4runner, i have no idea. Never owned or looked at G.skill much.
My question remains, lots more bandwidth with looser timings worth it? ... or would the OP article ddr4-2666 recommended target be ideal? with lower bandwidth it seems you can get tighter timings, and/or a C15 instead of C16 for your money.
http://www.ncix.com/detail/corsair-vengeance-lpx-32gb-4x8gb-ac-103069.htm
( spec: http://www.corsair.com/en-us/vengea...66mhz-c15-memory-kit-black-cmk32gx4m4a2666c15 )
My question remains, lots more bandwidth with looser timings worth it? ... or would the OP article ddr4-2666 recommended target be ideal? with lower bandwidth it seems you can get tighter timings, and/or a C15 instead of C16 for your money.
http://www.ncix.com/detail/corsair-vengeance-lpx-32gb-4x8gb-ac-103069.htm
( spec: http://www.corsair.com/en-us/vengea...66mhz-c15-memory-kit-black-cmk32gx4m4a2666c15 )
Commander Suzdal
2[H]4U
- Joined
- Jul 29, 2004
- Messages
- 2,134
As somebody finally upgrading from LGA775 yes, yes it is
Me too! My Yorkfield Q9550 has seen me through my always-2-to-4-years-behind gaming habits. Did Skyrim, almost through Fallout: New Vegas, have FarCry 3 on deck and thinking about CPU upgrade before I try to play it.
Thanks Dan,
I see that the EVGA FTW board has 7 Fan headers on it, but they also left off USB3.1 for some reason.
![Zarathustra[H]](https://cdn.hardforum.com/data/avatars/m/9/9298.jpg?1715876891){kind=avatar}
Zarathustra[H]
Extremely [H]
- Joined
- Oct 29, 2000
- Messages
- 39,073
I have an older Asus board (P9X79 WS) it has one CPU fan output, and a few case fan ones, but I have been using a 4 pin splitter, that essentially takes the PWM signal from the header and sends it to multiple fans, sends the RPM signal from one of them back, and pulls the power for all the fans from a molex plug.
This has worked well (much better than relying on Corsairs Link to control the fans). I manually set the fan control profile in the BIOS, and then if I need more detailed control use ASUS Fan Xpert in windows.
Corsair Link is currently in a pretty sad state, due in large part to shared control of firmware and software between Corsair and CoolIT, but apparently they have outsourced a new Link project to a third party to create a new version 4 from scratch, which if you are to believe the support guys on the Corsair forums is supposed to look pretty promising.
I'm coming to same conclusion.
I've been with Sandy since May, 2011
I can't come up with a real good reason to upgrade this i7-2600k 4.4Ghz o/c system to
the new processor.
The charts tell all when it comes to my main concern, gaming. THe charts
in this article were great but I found charts on another site that compared
real world performance using discreet GPU on higher resolution games so total
frame rates were as low as in 60's in various games they tested.
In other words GPU limited games...
Across CPU Sandy and up platforms the changes in frame rate were far less 'impressive' than the bar graphs here. .. Yah, 12 percent of 60 FPS is what? Right. No reason to blow 800 dollars for that.
One would be better off blowing that much green on a GTX980TI, which I did a couple
months ago, assuming upgrading a Sandy Bridge i5 2500k or better anyway.
The only department that the new Skylake beats pants off the Sandy for MY purposes,
is disk encryption with True Crypt or handling zip/rar archives.
In both cases I'm happy enough with Sandy doing that job so......
It is rather amazing the CPU I bought in 2011 is chugging along fine for me here 4 years
later and it appears will continue to do so for perhaps 4 more years if this is the
best Intel/AMD can do so far.
Not that I'm complaining that much ! I remember the upgrade cycles we had in 80's and 90's and turn of century. Every 2 years I was chasing another CPU back then.
Think of all the coin we are saving.
I think if I was heavily into video and or 3d modeling I'd have a different view
on Skylake vs Sandy.
Nenu
[H]ardened
- Joined
- Apr 28, 2007
- Messages
- 20,315
Finally got my 6600K upgrade running properly but killed a 2TB drive in the process.
Its a LOT snappier than my old 2500K.
I have the MSI Z170A Gaming Pro motherboard (the one with the LEDs that can do 16.5 million colours, although I'm yet to test that).
A good motherboard but is missing LLC.
It is brimming with all the other overclocking options you could want and as soon as you realise you must use the keyboard +/- keys to change many options, it will stop you swearing at it
It managed to get the 6600K to 4.6GHz with Cache at 4.5GHz (not tried the cache higher yet) @ 1.375V.
4.7GHz looks like it will need well over 1.4V so dont think I'll bother.
My Corsair Vengeance LPX 3000 CL15 is tough to clock with basic methods. I havent succeeded at stock voltage, I'll test more later. 3200MHz and CL14 both failed. I got partial success at 3100MHz.
It is stable at 1T instead of 2T.
Problems I hit...
...It wouldnt POST at first, LEDs indicated a ram problem so I used 1 stick and it worked ok from then on with all sticks.
...The Intel Management software slowed my PC to a crawl so uninstalled it. I have a newer version ready so hopefully that works better.
...I had it working from the motherboards Intel gfx out but when I tested it again later after putting my gfx card in, it blue screened every boot. I have installed the latest driver now but havent verified if the problem is fixed.
...I killed a 2TB hard drive. Some leads on the Corsair AX750 have 4 SATA power connections and the connections get stressed when the drives are close to each other. A bad power connection made a drive start clicking like mad and now it wont power up unless the SATA data cable is left off. I then reconnect it, reboot and its working but Windows moans the drive is dying. I now only use leads with 2 SATA power connectors.
...The original BIOS was V1.0 and was more difficult to get a stable overclock with.
...My 6600Ks temps are odd. The first 2 cores run 12C hotter than the last 2 under load.
When testing at 4.6GHz, The first 2 are close to 70C, the others are under 60C. Checked the paste and it was very well covered.
...USB 3.1 port performance with a USB 3.0 hard drive seems to be a fair bit slower than on my old motherboard. I've updated the main USB drivers but havent found new drivers for the USB 3.1 ports yet.
Ordinarily I would get 75MB/s+ but am now getting 50 to 75MB/s in general.
...Lack of LLC makes voltage droop under load. ie When I set it to 1.35V it is that at idle but with Prime95 it drops to under 1.30V. Not a biggy because Skylake gets too hot before voltage get high enough.
Performance is a lot snappier than my old [email protected] (on a Gigabyte P67A with 12GB DDR3-1600).
Windows 7-64 would take a short moment before loading a program, now its instant. (On a Samsung 840 Pro SSD)
Idle CPU use was around 5 to 12% total (mostly from just one core), now its 0 to 1% total.
Project Cars is no longer CPU bottlenecked, it has become incredibly smooth where there were noticeable jerks.
But PCars uses 90% CPU at times even when clocked, quite crazy.
I havent had chance to test anything else.
An expensive upgrade which would probably have been better served by getting a DDR3 motherboard and 16GB of low voltage DDR3.
That would have saved me about £100 from the £435 I spent but the motherboards werent available.
It feels like a fresh PC. I wasnt annoyed about its performance before but there is a distinct difference.
This is using the same Windows install which is a few years old.
I am very happy, it gives me the performance I was after in Project Cars.
Its a LOT snappier than my old 2500K.
I have the MSI Z170A Gaming Pro motherboard (the one with the LEDs that can do 16.5 million colours, although I'm yet to test that).
A good motherboard but is missing LLC.
It is brimming with all the other overclocking options you could want and as soon as you realise you must use the keyboard +/- keys to change many options, it will stop you swearing at it
It managed to get the 6600K to 4.6GHz with Cache at 4.5GHz (not tried the cache higher yet) @ 1.375V.
4.7GHz looks like it will need well over 1.4V so dont think I'll bother.
My Corsair Vengeance LPX 3000 CL15 is tough to clock with basic methods. I havent succeeded at stock voltage, I'll test more later. 3200MHz and CL14 both failed. I got partial success at 3100MHz.
It is stable at 1T instead of 2T.
Problems I hit...
...It wouldnt POST at first, LEDs indicated a ram problem so I used 1 stick and it worked ok from then on with all sticks.
...The Intel Management software slowed my PC to a crawl so uninstalled it. I have a newer version ready so hopefully that works better.
...I had it working from the motherboards Intel gfx out but when I tested it again later after putting my gfx card in, it blue screened every boot. I have installed the latest driver now but havent verified if the problem is fixed.
...I killed a 2TB hard drive. Some leads on the Corsair AX750 have 4 SATA power connections and the connections get stressed when the drives are close to each other. A bad power connection made a drive start clicking like mad and now it wont power up unless the SATA data cable is left off. I then reconnect it, reboot and its working but Windows moans the drive is dying. I now only use leads with 2 SATA power connectors.
...The original BIOS was V1.0 and was more difficult to get a stable overclock with.
...My 6600Ks temps are odd. The first 2 cores run 12C hotter than the last 2 under load.
When testing at 4.6GHz, The first 2 are close to 70C, the others are under 60C. Checked the paste and it was very well covered.
...USB 3.1 port performance with a USB 3.0 hard drive seems to be a fair bit slower than on my old motherboard. I've updated the main USB drivers but havent found new drivers for the USB 3.1 ports yet.
Ordinarily I would get 75MB/s+ but am now getting 50 to 75MB/s in general.
...Lack of LLC makes voltage droop under load. ie When I set it to 1.35V it is that at idle but with Prime95 it drops to under 1.30V. Not a biggy because Skylake gets too hot before voltage get high enough.
Performance is a lot snappier than my old [email protected] (on a Gigabyte P67A with 12GB DDR3-1600).
Windows 7-64 would take a short moment before loading a program, now its instant. (On a Samsung 840 Pro SSD)
Idle CPU use was around 5 to 12% total (mostly from just one core), now its 0 to 1% total.
Project Cars is no longer CPU bottlenecked, it has become incredibly smooth where there were noticeable jerks.
But PCars uses 90% CPU at times even when clocked, quite crazy.
I havent had chance to test anything else.
An expensive upgrade which would probably have been better served by getting a DDR3 motherboard and 16GB of low voltage DDR3.
That would have saved me about £100 from the £435 I spent but the motherboards werent available.
It feels like a fresh PC. I wasnt annoyed about its performance before but there is a distinct difference.
This is using the same Windows install which is a few years old.
I am very happy, it gives me the performance I was after in Project Cars.
obiwansotti
Gawd
- Joined
- Dec 13, 2002
- Messages
- 1,003
did you wipe re-install windows or just swap out the hardware.
Windows rot is a real thing. Perceived windows performance improvement could just be a clean install.
Glad to hear pCars is working better for you, that and witcher 3 are the games that I feel like could use more CPU for my GTX 970.
Windows rot is a real thing. Perceived windows performance improvement could just be a clean install.
Glad to hear pCars is working better for you, that and witcher 3 are the games that I feel like could use more CPU for my GTX 970.
Nenu
[H]ardened
- Joined
- Apr 28, 2007
- Messages
- 20,315
Taa
Its a 2 years old Windows install (mentioned in 2nd to last line). I uninstalled a few drivers and then just plugged the new board in.
Yeah Witcher 3 will be my next test.
Its a 2 years old Windows install (mentioned in 2nd to last line). I uninstalled a few drivers and then just plugged the new board in.
Yeah Witcher 3 will be my next test.
obiwansotti
Gawd
- Joined
- Dec 13, 2002
- Messages
- 1,003
I've been having a great time watching folks prostulate and ring there hands over Skylake and Win10.
loaded win10 about a week before official RTW.
This 5820K v 6700K has me all twisted though.
Same, but I have convinced myself that whats really bugging me is that I could go buy a 5820k today and I cant buy the 6700k which makes me want to buy whats available, at least that is my issue.
I am almost positive I will wait on the 6700k now and my I7-920 can become my new Plex server. I like to build PCs and I been waiting since 2009 to build a new one I think its time.
Anyone waiting on windows 10, don't wait any longer it works great.
4RunnerTE
Limp Gawd
- Joined
- Aug 12, 2015
- Messages
- 222
Depends on your needs, but honestly can't go wrong with either. Unless you are really tight on $$ and opting for a budget z170 board in which case you will probably come out ~$50. That may or may not matter to some.
Then again if you need the extra cores it's also a no-brainer. If you're tired of waiting (like me), then yes, you become all twisted.
I know it will break something I use for reviews. That's why I haven't switched. But I hate 8.1 and would love to load the new OS.
For any of you guys who are on the fence of 6700k vs 5820k, I would go with the 5820k. There is a saying of "when in doubt there is no doubt". Unless you use compiling apps on a continuous basis that really require every last ounce of IPC strength, 50% more cores is a better upgrade. IMHO That is of course unless you don't need the extra cores, but if you're on the fence something tells me you guys could use them.
Just my opinion.
Just my opinion.
This may sound like a stupid question, but I have been out of the processor hardware loop for awhile. Would a i7 6700k be an upgrade from my modest overclocked i7 3930k? I am starting to get the upgrade itch and i was just curious, since I only game now and dont do much converting or editing.
The CPU isn't world's faster than the 3930K depending on how you've got that overclocked. The platform on the other hand has some significant upgrades.
Nothing silly about that IMO, my current MSI P67 board was kinda limited in that regard and I didn't realize it until a while after purchase (not all headers are controllable, only DC, etc)...
I for one actually prefer using headers that you can automate via mobo software or Speedfan vs acontroller I have to manually tune or tie to thermal probes.
Little things like that or having two internal USB 3.0 headers (for card reader + case ports) is why I went back to an ASUS this go around. I was pretty satisfied with the MSI nonetheless.
Same here. I monitor everything just the same, but prefer the automated cooling process.
If there must be an external controller then I like the programmable ones.
I never liked wiring up fan controllers and tuning them. I had one for awhile and quickly abandoned it. I'm pretty satisfied with what's onboard ASUS motherboards right now. The others haven't been as good in this regard. Hopefully this generation's Z170 offers from MSI and GIGABYTE are better in this regard. (Which I believe, is the case for MSI.)
Dan,
This is probably a silly question, but are there any issues with controlling 2 pwm fans off the same header with a Y-splitter? All but one of my fans are installed in pairs, 2 CPU, 2 front, 2 top and 1 rear. Noctua is nice enough to provide a splitter with every fan so I was thinking it made more sense to control them by zone rather than individual fan.
I honestly don't know. I had a ton of fan headers on some ABIT boards burn out doing that back in the day. So I've never put more than one fan on a single header since. Raja from ASUS might be able to answer that question.
Yeah I used to use Speedfan until Asus Fan Xpert came out, it's a perfectly good tool. I even find the bios curve settings good enough.
All the Z170 motherboards offer virtually everything you had in software with the Z97 boards in the UEFI now. The fan ramp up and ramp down delays were added to the UEFI this generation. Previously that was a TUF series only feature. You can also remap the headers in UEFI now. The only thing you can't do is rename them to something custom like you can in Fan XPert.
Good to know
Uhhh actually the opposite is true, LLAPI's take the weight off the CPU and onto the GPU thus making them more GPU bound than CPU bound, that is why CPU is gonna matter less, and it has been their entire point.
It depends on how intense the developer intends to the scenes to be. If there are a lot of objects (they are no longer limited remember) then the CPU will matter a lot. If the object count doesn't increase much then the CPU will mean very little.
A lot of stress was taken away from both the GPU and the CPU with techniques added from DX12 and Vulcan. But it doesn't mean the developer can't go buck-wild crazy either.
A lot of stress was taken away from both the GPU and the CPU with techniques added from DX12 and Vulcan. But it doesn't mean the developer can't go buck-wild crazy either.
With only 16 PCIe CPU lanes for the GPU, forget using a NVMe SSD with these systems since either your GPU is reduced to x8 or the SSD must share the DMI bottleneck with all the peripherals.
Your sort of right. Dropping the GPU to 8x PCIe 3.0 isn't really that big of a deal. Secondly, the SSD does have to go across the DMI 3.0 bus but at 32Gbps it still leaves a fair amount of bandwidth for the other devices. Keep in mind that none of these will saturate the DMI bus consistently. You can do it in benchmark tests if your just trying to for the sake of doing it. Rarely will you be maxing out the bandwidth of a gigabit NIC, USB 3.0 or 3.1 ports, nor even the SSD. I mention the issue of DMI bandwidth primarily for configurations using two or more NVMe SSD's in RAID 0 across the DMI bus. In a single drive scenario I wouldn't worry too much about it.
With only 16 PCIe CPU lanes for the GPU, forget using a NVMe SSD with these systems since either your GPU is reduced to x8 or the SSD must share the DMI bottleneck with all the peripherals.
Yeah, I don't think it works that way. I believe the CPU has 20 total lanes. The GPU get's 16 (or 2 get 8 each), but the PCIe x4 slot uses on the other 4 not allocated to the GPU.
Please correct me if I'm wrong (I didn't go look this up, just what I remember reading somewhere recently).
EDIT: Hmm, so I'm reading that the CPU has 16 lanes, but the PCH has 20? Now I'm just confused
EDIT: Hmm, so I'm reading that the CPU has 16 lanes, but the PCH has 20? Now I'm just confused
The CPU has only 16 PCIe 3.0 lanes, the Z170 chipset has 20 PCIe 3.0 lanes but all those PCIe lanes plus the SATA, USB , ethernet have to go through the DMI connection between the CPU and chipset.
i7-5820K and X99 is much cleaner way to go for a little more money.
Agreed the GPU going from 16x to 8x is not that much of a performance hit, but the DMI bus will be a bottleneck for SATA, ethernet, USB and NVMe SSDs together. The DMI 3,0 link is equivalent bandwidth to just 4x PCIe lanes. The new Samsung NVMe drives do 5.5GB/sec reads. That pretty much uses up your DMI 3.0 bandwidth.
SilverSliver
[H]F Junkie
- Joined
- Feb 23, 2004
- Messages
- 12,555
You guys think I will see much improvement at 4k with SLI 980 Ti over my 2600k?
Nenu
[H]ardened
- Joined
- Apr 28, 2007
- Messages
- 20,315
Doubtful.
The chance that RAID drives will bottleneck Z170 DMI is slim because its rare they achieve max transfer rate.