|
I like the LGA setup, personally, because removal of the HSF won't yank the drat chip out of the socket half the time.
|
#
?
Jul 19, 2015 04:31
|
|
|
|
| # ? Apr 18, 2024 13:34 |
|
|
JnnyThndrs posted:I like the LGA setup, personally, because removal of the HSF won't yank the drat chip out of the socket half the time. Except they could have kept PGA and just added a simple over-chip retention clip as part of the assembly and accomplished the exact same thing. Like I said before, I was petrified getting my 2500K locked in years ago because it was my first time dicking around with Intel's new LGA setup, and I could not fathom the amount of force I had to exert on the bar to finally get it down and locked. I thought the bar was going to break or bend, or worse, something was going to bend/break in the socket itself.
|
|
#
?
Jul 19, 2015 04:36
|
|
|
HalloKitty posted:Haswell. Haswell's got the FIVR.
|
|
#
?
Jul 19, 2015 04:46
|
|
|
Given I have a 2600k at 4.3Ghz is Skylake a good point to upgrade or would it be worthwhile to wait until Cannonlake? My choice in the upcoming 12 months is either getting a new cpu/mobo/ddr4 or upgrading my 780 when nvidia releases their new line.
|
|
#
?
Jul 19, 2015 21:59
|
|
|
JnnyThndrs posted:I like the LGA setup, personally, because removal of the HSF won't yank the drat chip out of the socket half the time. Thanks for reminding me of the horror I felt when I managed to do just that with a Socket 939 chip years ago.
|
|
#
?
Jul 19, 2015 22:21
|
|
Haeleus posted:Given I have a 2600k at 4.3Ghz is Skylake a good point to upgrade or would it be worthwhile to wait until Cannonlake? My choice in the upcoming 12 months is either getting a new cpu/mobo/ddr4 or upgrading my 780 when nvidia releases their new line. I'd wait if you don't have a good reason to upgrade, the leaked benches show Skylake as another very minor bump for gaming unless you are using the IGP.
|
|
|
#
?
Jul 19, 2015 23:30
|
|
|
Also, Intel mentioned Broadwell-E recently so I guess in light of them pushing Cannonlake back, I guess they've decided to come out with that now versus putting more effort into Skylake-E as originally rumored. I should note that it was WCCFTech that leaked the rumor. Supposedly B-E will use Socket 2011-3 again, so it gives you the ability to watch for deals on current-gen boards, since I'm sure all it'll require to make them B-E-ready would be a BIOS update (one can hope).
|
|
#
?
Jul 20, 2015 00:33
|
|
|
Haeleus posted:Given I have a 2600k at 4.3Ghz
|
|
#
?
Jul 20, 2015 00:48
|
|
|
Are there any online stores that buy large DDR4 ram kits and then split them up to sell, or is the answer probably craigslist?
|
|
#
?
Jul 20, 2015 01:04
|
|
|
Josh Lyman posted:Just to be clear, when people say this, they're referring to the CPU being able to run at that multiplier under full load but it's not actually running at that speed all the time, right? Yeah, most of the time it's clocked down and only 'revs up' under load - not even all the cores will ramp up to the top multi, unless you're running Prime95 or an encoder or something.
|
|
#
?
Jul 20, 2015 01:09
|
|
|
Josh Lyman posted:Just to be clear, when people say this, they're referring to the CPU being able to run at that multiplier under full load but it's not actually running at that speed all the time, right? Yeah its the max sustained turbo multiplier, and when there is no load it idles back to 1.6GHz. BTW I was playing with the idea of toying around a used Q9550 just for kicks, until I remembered a brand new stock G3258 simply roflstomps it with something like a ~50% IPC advantage with all the advantages of a modern chipset at around the same price. Poor second hand sellers these days.
|
|
#
?
Jul 20, 2015 01:12
|
|
|
Josh Lyman posted:Just to be clear, when people say this, they're referring to the CPU being able to run at that multiplier under full load but it's not actually running at that speed all the time, right? I've actually setup BIOS to run my cpu @ 4.3 Ghz constantly (aka 43 multiplier with turbo mode constantly activated), but normally yes this is turbo mode with the CPU throttling back to 1.6 when idle.
|
|
#
?
Jul 20, 2015 02:57
|
|
|
AVeryLargeRadish posted:I'd wait if you don't have a good reason to upgrade, the leaked benches show Skylake as another very minor bump for gaming unless you are using the IGP. So no reason to upgrade my 2500K either?
|
|
#
?
Jul 20, 2015 03:11
|
|
Drunk Badger posted:So no reason to upgrade my 2500K either? For the most part a OCed 2500k will be just as fast as the newer chips at the same clocks. But it does depend on your workload so you might want to read some reviews and see if newer chips have anything significant to offer you.
|
|
|
#
?
Jul 20, 2015 03:23
|
|
|
AVeryLargeRadish posted:For the most part a OCed 2500k will be just as fast as the newer chips at the same clocks. But it does depend on your workload so you might want to read some reviews and see if newer chips have anything significant to offer you. From the initial Skylake benchmarks there isn't much of an imperative reason to upgrade. Performance gains are roughly 5-10% and SATA-Express with NVMe isn't that much of a boost either. My plan is to try and wait for whatever's after Cannon Lake which won't be until 2018-2019.
|
|
#
?
Jul 20, 2015 03:33
|
|
|
Haeleus posted:I've actually setup BIOS to run my cpu @ 4.3 Ghz constantly (aka 43 multiplier with turbo mode constantly activated), but normally yes this is turbo mode with the CPU throttling back to 1.6 when idle. Curious why you would do this?
|
|
#
?
Jul 20, 2015 03:51
|
|
|
Sidesaddle Cavalry posted:Are there any online stores that buy large DDR4 ram kits and then split them up to sell, or is the answer probably craigslist? Massdrop does something similar to this, but they don't do a lot of computer parts really.
|
|
#
?
Jul 20, 2015 05:18
|
|
|
Massdrop looks like they list by full official kits only and I was hoping there would be somebody who literally takes them out of the package and splits them up.
|
|
#
?
Jul 20, 2015 05:38
|
|
|
Josh Lyman posted:Just to be clear, when people say this, they're referring to the CPU being able to run at that multiplier under full load but it's not actually running at that speed all the time, right? On my 2500k @ 4.8 ghz, once every 9 months or so I'll get a BSOD when I'm just idling / listening to winamp at desktop. I believe I've tracked it down to the C1E, C3, and the C6 states. Under Prime95 / full load I've never had it crash ever, but sometimes when it drops into idle mode it's possible to very briefly get the "whoa not enough vcore!" BSOD. So in some cases it's a good idea to turn those states off. Interestingly it's stable at 5.0 ghz but because it goes over 90c it gave a lower score in Cinbench.  That reminds me, I've been using a H100 AIO for about 3 years now, and it might be my imagination but temps seem to be slightly higher than when I first brought it. I wish there was a way of checking the fluid in it, but then that wouldn't be a closed loop system. I've got my eye on the newer H110i GT but only if I have to replace the H100.
|
|
#
?
Jul 20, 2015 09:51
|
|
|
Thanks guys. I feel much better about only pushing my 3570K to 4.2GHz. 
|
|
#
?
Jul 20, 2015 10:37
|
|
|
Josh Lyman posted:Thanks guys. I feel much better about only pushing my 3570K to 4.2GHz. It's enough, and it'll be stable, so I'd be pleased with it. Have my 2500K @ 4.4, don't really feel the need to potentially make the system less stable, and with the slowing rate of performance increases, I'd want it to last for a longer time than I had anticipated.
|
|
#
?
Jul 20, 2015 13:03
|
|
|
HalloKitty posted:It's enough, and it'll be stable, so I'd be pleased with it. Have my 2500K @ 4.4, don't really feel the need to potentially make the system less stable, and with the slowing rate of performance increases, I'd want it to last for a longer time than I had anticipated. This. My 2500k is at 4.2GHz and never crashes. Sure I could push it some more, but I want stability and use of the idle states where it spends 98% of the time it is on. I got a new Kill-A-Watt and did some testing over the the weekend my box with a GTX 680, 3 SSDs, 1 HDD, and 16GB of RAM idles around 88W and goes to 115W or so with a light to moderate load like Photoshop or Lightroom. Firing up GTA V that completely maxes my system gets me to 340W. Prime95 gave me a load of 225W, so really the 680 was only using about 115W. I will gladly give up a few hundred Megahertz for a 2/3 reduction in power draw.
|
|
#
?
Jul 20, 2015 13:44
|
|
|
wipeout posted:Curious why you would do this? Maybe he only boots the PC up when he has to do fairly long-running heavy workload jobs and the CPU trying to throttle up and down at perceived dips in activity hurts his performance on those jobs? No idea what kinda job that would be, but in my OS class we definitely fooled the CPU's throttling quite a lot by deliberately writing programs that just churned away doing whatever and then blipped to 0 utilization for a few MS and then went back to heavy work. In those situations, the CPU's throttling more than doubled how long it actually took to get back to working at full capacity.
|
|
#
?
Jul 20, 2015 14:33
|
|
|
Combat Pretzel posted:Talking about pressure optimized airflows, is there a reason why CPU fans don't have stators? Well, there's that single fan model from Noctua that has some, but that's about it. I'd venture to guess the fine parallel-plate structure of the heatsink does a fantastic job of promoting laminar flow. Edit: actually, the leading edges are so drat close together -- I really have no clue. Retracted. Potato Salad fucked around with this message at 16:31 on Jul 20, 2015 |
|
#
?
Jul 20, 2015 16:27
|
|
|
Is there anything on the horizon to replace Silvermont for SoC storage servers? Newegg.ca is selling octocore Avotons for 20% off and my bonus is burning a hole in my wallet.
|
|
#
?
Jul 20, 2015 17:38
|
|
|
ElehemEare posted:Is there anything on the horizon to replace Silvermont for SoC storage servers? Newegg.ca is selling octocore Avotons for 20% off and my bonus is burning a hole in my wallet. How deep are your pockets and do you want onboard 10GigE? http://www.anandtech.com/show/9185/intel-xeon-d-review-performance-per-watt-server-soc-champion
|
|
#
?
Jul 20, 2015 17:40
|
|
|
wipeout posted:Curious why you would do this? I just did it over a year ago as I was experimenting with OCing and thought that having the CPU change clock speed would risk stuttering in games (literally all I care about performance wise). I guess in retrospect that is unnecessary; since I don't remember what I changed in BIOS I'll just reset to factory.
|
|
#
?
Jul 20, 2015 18:16
|
|
|
That could sometimes happen on socket 775 boards, or they'd just get hella unstable when changing states like my Q6600 box that had to be hard set at 3.6ghz. Any modern CPU should handle dynamic overclocking fine unless the board's garbage or dvid is set incorrectly though, so you're just stressing the chip and wasting electricity really.
|
|
#
?
Jul 20, 2015 19:35
|
|
|
Are there any thunderbolt pci-e cards available that don't need a thunderbolt header on the motherboard? Or are thunderbolt in a weird limbo of not being successful but backed by billions of dollars.
|
|
#
?
Jul 20, 2015 20:39
|
|
|
Twerk from Home posted:How deep are your pockets and do you want onboard 10GigE? Medium and no. Guess the C2750 is good enough for me.
|
|
#
?
Jul 20, 2015 21:54
|
|
|
Boiled Water posted:Are there any thunderbolt pci-e cards available that don't need a thunderbolt header on the motherboard? Or are thunderbolt in a weird limbo of not being successful but backed by billions of dollars.
|
|
#
?
Jul 20, 2015 21:54
|
|
|
Combat Pretzel posted:Weird limbo. Plus some bullshit between Intel and Apple. Maybe try eBay? There are lots of grey-market electronics that get sold there. Don't expect a warranty or it to work terribly well though.
|
|
#
?
Jul 21, 2015 00:07
|
|
|
Apparently it's possible to use other USB headers and whatnot if the mainboard manufacturer can be bothered for a BIOS update, but you need something to plug into. HP did it for some older workstations. Their little addon cards are 175$ lol. Don't get your hopes up in the low end segment.
|
|
#
?
Jul 21, 2015 00:32
|
|
|
Skandranon posted:Maybe try eBay? There are lots of grey-market electronics that get sold there. Don't expect a warranty or it to work terribly well though.
|
|
#
?
Jul 21, 2015 00:45
|
|
|
cisco privilege posted:That could sometimes happen on socket 775 boards, or they'd just get hella unstable when changing states like my Q6600 box that had to be hard set at 3.6ghz. Any modern CPU should handle dynamic overclocking fine unless the board's garbage or dvid is set incorrectly though, so you're just stressing the chip and wasting electricity really.
|
|
#
?
Jul 21, 2015 01:52
|
|
|
http://www.kitguru.net/components/cpu/anton-shilov/intel-core-i7-6700k-skylake-overclocked-to-5-20ghz-with-air-cooling/ Is this news to anyone? Sorry I wasn't keeping up here but this is not at all what I was expecting.
|
|
#
?
Jul 21, 2015 02:51
|
|
|
THE DOG HOUSE posted:http://www.kitguru.net/components/cpu/anton-shilov/intel-core-i7-6700k-skylake-overclocked-to-5-20ghz-with-air-cooling/ Maybe with the FIVR off-chip this time though, we can get something... who knows?
|
|
#
?
Jul 21, 2015 03:03
|
|
|
LiquidRain posted:We heard this before with Devil's Canyon though. Intel was going so far as saying that if you couldn't get 5GHz on air you were doing something wrong. (practically nobody got above 4.8) True. Still, its better than the opposite I guess. And this is a bit more positive (if true) since its not intel based info. Frankly I thought the OC headroom was just going to decline a bit more this time around but, shows how far out of the loop ive fallen
|
|
#
?
Jul 21, 2015 03:04
|
|
|
LiquidRain posted:We heard this before with Devil's Canyon though. Intel was going so far as saying that if you couldn't get 5GHz on air you were doing something wrong. (practically nobody got above 4.8) Well, you get more TDP flexibility on the part of the CPU and iGPU between the higher TDP on the top end pieces and the 5-10(?) ish watts saved by moving the voltage regulators back off the package. I'm sorta surprised they did that since the trend for so long has been increasing integration and simplification of packaging stuff drive by the THIN SMALL LIGHT mania for laptops and tablets.
|
|
#
?
Jul 21, 2015 03:22
|
|
|
|
| # ? Apr 18, 2024 13:34 |
|
|
THE DOG HOUSE posted:True. Still, its better than the opposite I guess. And this is a bit more positive (if true) since its not intel based info. One positive sign about Skylake is that the non-K chips have a 65W TDP and very similar clocks to the -K parts. This indicates to me that there's a good amount of TDP headroom they're leaving users on the -K chips, with their 95W TDP.
|
|
#
?
Jul 21, 2015 03:43
|
|