|
redeyes posted:Just got an ASRock X99 Fatality Killer system. Bluescreening like crazy. First issue, the BIOS/UEFI battery, totally dead. System wont turn on without a good battery. Next, bluescreen instantly playing youtube. That turned out to be the Killer gaming NIC. Just turned off that stupid thing and use the built in Intel NIC. Last, random bluescreens pointing to the Windows kernel which usually means Memory. System uses G.Skill DDR4 2666 stuff. I upped the voltage to 1.25v from 1.20 and that seems to have stabilized the thing. G.Skill poo poo sucks and I don't trust it. Yes, when socket 1151 came out (we dont do a whole lot of x99 builds) we had lots of issues with gskill memory and asrock boards. Weve had good luck with the cheap corsair lpx ram, kingston hyperx, and the totally not fancy at all crucial ddr4 https://www.amazon.com/Corsair-Veng...rds=corsair+lpx
|
#
?
Jun 16, 2016 15:21
|
|
|
|
| # ? May 11, 2024 08:08 |
|
|
Don Lapre posted:Yes, when socket 1151 came out (we dont do a whole lot of x99 builds) we had lots of issues with gskill memory and asrock boards. I'm running Crucial DDR4 2133 non overclocking modules and they are rock solid. I've had bad experiences with G.Skill for years. At this point they are the OCZ of memory for me. 
|
|
#
?
Jun 16, 2016 15:32
|
|
|
Mushkin has been proven for me as well. Only had 1 bad SO-DIMM that was DOA that they replaced without a fuss out of years of kits.
|
|
#
?
Jun 16, 2016 16:01
|
|
|
necrobobsledder posted:This was back in 2002 and Intel hadn't invested much in schedulers compared to waging the clock speed war.  Yeah, ramping up the clock speed works without any appreciable effort on any other component.You're confusing the Pentium 4's double-pumped ALU (which shares a passing resemblance to DDR's clocking) with SMT, which was coincidentally implemented on the P4. That implementation did not persist through the Core lines.
|
|
#
?
Jun 16, 2016 17:25
|
|
|
The only brands I have not seen fail at this point are ASRock and Crucial. Literally every other motherboard/video card/RAM maker has failed on me personally at one point or another over the decades.
|
|
#
?
Jun 17, 2016 19:33
|
|
|
Kind of hilarious, considering that ASRock was the cheap brand label, before they spun off from Asus. I'm extremely satisfied with my X99 Extreme4 and E3C226D4I mainboards of theirs.
|
|
#
?
Jun 17, 2016 22:52
|
|
|
My Asrock Z170 Extreme 4 refused to clock my Gskill 3200 RAM anything above about 2400, until the recent V3.20 BIOS. Now it does 3200 without issue. At a guess the compatibility has been improved a bunch across other boards with bleeding edge BIOS (It should've worked in the first place 6 months ago though)
|
|
#
?
Jun 18, 2016 04:53
|
|
|
You guys are making me happy I spent the extra 10 bucks on Corsair.
|
|
#
?
Jun 18, 2016 05:37
|
|
|
LiquidRain posted:You guys are making me happy I spent the extra 10 bucks on Corsair. I've had Corsair fail on me before, but it was a 4x4GB kit. I think that was down to memory controllers still by and large being dogshit at dealing with fully-populated DIMM slots. Ever since going to 2x8GB I've been rock solid. Not going to love life when Skylake-E rolls around, though - six-channel RAM. I kind of enjoy the idea of only having two DIMMs - it makes finding 'the dying stick' so much easier.
|
|
#
?
Jun 18, 2016 06:43
|
|
BIG HEADLINE posted:I've had Corsair fail on me before, but it was a 4x4GB kit. I think that was down to memory controllers still by and large being dogshit at dealing with fully-populated DIMM slots. Ever since going to 2x8GB I've been rock solid. Not going to love life when Skylake-E rolls around, though - six-channel RAM. I kind of enjoy the idea of only having two DIMMs - it makes finding 'the dying stick' so much easier. Yeah, it's probably the memory controller, hell I was reading a review for the ASRock Z170M OC Formula which is supposed to be able to run RAM at over 4500MHz and they only put two RAM slots on it because they found that putting four on was just stupid because it's impossible to run any more than two DIMMs at those sorts of speeds. Though I find the idea of 4500MHz RAM pretty hilarious, I guess the highest number is all that matters to some people.
|
|
|
#
?
Jun 18, 2016 07:16
|
|
|
Still oscillating between a 5820k and a 6700k. You know what burns me up? SATA Express. Who gives a poo poo that your Z170 motherboard has three SATA Express ports? poo poo's been dead since 2014, nobody cares. There do not even exist (to my knowledge) any devices you can plug into it and there probably never will be. Paul MaudDib fucked around with this message at 07:20 on Jun 18, 2016 |
|
#
?
Jun 18, 2016 07:17
|
|
Paul MaudDib posted:Still oscillating between a 5820k and a 6700k. Some of the nicer mobos come with a 5.25" bay adapter that runs off of SATA Express and adds a couple USB 3.1 Gen 2 Type A and C ports to the front of your computer, I can see that as somewhat useful but that is about the only use I have seen for them recently.
|
|
|
#
?
Jun 18, 2016 07:21
|
|
|
So I'm looking at options for a FreeNAS build using a skylake chipset that supports ECC, C236 in this case. Why is it most of the motherboards use unbuffered ECC (DDR4), and why is it that it seems that registered is much more plentiful and cheaper? I was all set to get the C236M from MSI but I looked at the limited selection of unbuffered ECC DDR4 and cringed. Is it just that not much stuff used unbuffered DDR4 previously and the product pipeline hasn't gotten filled up yet?
|
|
#
?
Jun 18, 2016 07:24
|
|
|
Paul MaudDib posted:You know what burns me up? SATA Express. Who gives a poo poo that your Z170 motherboard has three SATA Express ports? poo poo's been dead since 2014, nobody cares. There do not even exist (to my knowledge) any devices you can plug into it and there probably never will be. Especially when the head of Intel's storage division has gone on the record and said: “I’m not planning to offer SSDs in SATA Express,” Leszinske said of the neglected interface. “I don’t need another form factor or interface. It just complicates the product line.” Of course, in the very same article he claims that "SATA drives will become passe" and that the ~futcha~ is in *PCIe* drives. Granted, that statement could include M.2 drives connected to the PCIe interface, but still... Article in question: http://www.pcworld.com/article/3052391/storage/storage-on-steroids-intel-bets-on-pcie-ssds-and-optane-storage-as-fast-as-ram.html I'd not heard the first Optane drives would be offered in PCIe. I guess on Kaby Lake/Haswell-E+ you'll have a legitimate use for a second x16 port. BIG HEADLINE fucked around with this message at 09:21 on Jun 18, 2016 |
|
#
?
Jun 18, 2016 08:56
|
|
|
Paul MaudDib posted:Still oscillating between a 5820k and a 6700k.
|
|
#
?
Jun 18, 2016 11:58
|
|
|
Paul MaudDib posted:Still oscillating between a 5820k and a 6700k. I did this for a good while, then went for skylake. Mostly because of the improved per core performance, also a socket that should hopefully support a couple more chips. I hope I don't end up regretting it if dx12 / vulkan games performance gets a core boner.
|
|
#
?
Jun 18, 2016 12:53
|
|
|
Just installed the Hyper 212 EVO cooler for my good ol' 2500k. Currently not overclocked, idle temps are about 30C (ambient maybe about 20C). Am I good to go? Not sure if I should wait a while for the paste to spread or whatever before attempting to overclock it/do stress testing.
|
|
#
?
Jun 18, 2016 14:29
|
|
|
Paul MaudDib posted:Still oscillating between a 5820k and a 6700k. The answer is clearly 6800k.
|
|
#
?
Jun 18, 2016 15:06
|
|
|
priznat posted:So I'm looking at options for a FreeNAS build using a skylake chipset that supports ECC, C236 in this case. Registered memory is used by most servers since you need it to support more than the standard 2 DIMMs per channel, I believe. Most systems that support ECC also support (or require) registered memory, and most systems that don't support registered memory don't support ECC either. Unbuffered ECC is pretty much exclusively the domain of small single-socket servers like the one you're building, and so volume is lower and prices are higher. Recently I went to pick up some unbuffered ECC DDR3 for an older x58 system, and found that I could get it on eBay for a third of the cost of new. DDR4 probably isn't that cheap yet because no one is really liquidating any DDR4 systems yet, but it couldn't hurt to try. sout posted:Just installed the Hyper 212 EVO cooler for my good ol' 2500k. Yes. I recently repasted my 2500K (@4.4GHz, stock voltage) sitting under a Hyper 212+ with Tuniq TX-2, which I think has fairly standard properties for thermal grease, and I saw a substantial improvement over my old temps immediately but not much change from the paste curing over several days. I'm getting around 30-35 idle and barely 60 at load, so if you see 30 idle you should be fine. Eletriarnation fucked around with this message at 15:19 on Jun 18, 2016 |
|
#
?
Jun 18, 2016 15:08
|
|
|
Haven't done burn testing but I'm very happy with the CPU temps in-game with this OC profile, in the 40s and 50s Celsius (not necessarily CPU intensive games, should test that more later.) One weird thing, and this has been the case on my old cooler as well, but my motherboard seems to give false readings for some of my fans' RPMs (showing up as red/dangerous when the temperatures are totally normal). I disabled the monitor for this so it no longer stops me from launching windows to warn me of this etc, but it would still be nice to know my actual fan speeds. Is this just a problem with my motherboard? I can't imagine there's any quick fix to this and I can deal with it, still weird.
|
|
#
?
Jun 18, 2016 18:18
|
|
|
Hard to believe used AMD 1100Ts are still moving for ~$180. The AMD diehards that are rationalizing keeping or running a slow and ancient space heater drawing ~80W more power at load than Haswell/Skylake and a ton more OCed on ancient chipsets while feeling  and saying Intel is a ripoff is hilarious to say the least.Same goes for the X58 4GHz OC crowd about how their systems is nearly drawing extra 100W idle and 200W load power over Haswell/Skylake while saying Intel hasn't put out compelling chips to upgrade yet. If they have dumped X58 for Sandy Bridge when it came out they would have saved ~$150 in electricity bills @ 4 hours full load per day alone by now, not to mention the resale value of their parts. Palladium fucked around with this message at 05:27 on Jun 19, 2016 |
|
#
?
Jun 19, 2016 05:13
|
|
|
Palladium posted:they would have saved ~$150 in electricity bills @ 4 hours full load per day alone by now, not to mention the resale value of their parts.
|
|
#
?
Jun 19, 2016 05:36
|
|
|
SourKraut posted:So depending on what they would have gotten for the parts, maybe it'd have been break even since it'd have cost $300-400 easily for a i5-2500K and motherboard. The smart crowd would have easily made a profit by upgrading to a 2500K/cheap P67 combo at 2011 resale prices while getting higher performance all around and power savings gravy, while the not-so-smart ones are still mulling their heads over the sunk costs they paid for their X58 setup.
|
|
#
?
Jun 19, 2016 05:49
|
|
|
Palladium posted:The smart crowd would have easily made a profit by upgrading to a 2500K/cheap P67 combo at 2011 resale prices while getting higher performance all around and power savings gravy, while the not-so-smart ones are still mulling their heads over the sunk costs they paid for their X58 setup. But part of the whole argument is couched in 20/20 hindsight. No one (outside of Intel) at the time knew that performance would stagnate as it has, and that Sandy Bridge would still be as viable as it is today. So the majority of people would have argued to hold on to Nehalem/x58 for another generation or two and make the upgrade really worth it. Instead, Intel went a different direction to where we are today and so the above argument looks better in hindsight.
|
|
#
?
Jun 19, 2016 06:59
|
|
|
SourKraut posted:But part of the whole argument is couched in 20/20 hindsight. No one (outside of Intel) at the time knew that performance would stagnate as it has, and that Sandy Bridge would still be as viable as it is today. So the majority of people would have argued to hold on to Nehalem/x58 for another generation or two and make the upgrade really worth it. Instead, Intel went a different direction to where we are today and so the above argument looks better in hindsight. Huh what hindsight? Even when acting on information available at 2011 selling X58 for a new better performing Sandy Bridge platform while breaking even or better was already the rational choice back then. The question of whether X58 is still good enough in 2011 to wait for one or two more generations becomes irrelevant.
|
|
#
?
Jun 19, 2016 07:19
|
|
|
Has Skull Canyon been tested with an eGPU over Thunderbolt yet, and if so, was its iGPU memory found to be used as true L4 cache? Kind of looking forward to see if this design concept will keep carrying on to Kaby Lake (where Optane support begins) in some form.
Sidesaddle Cavalry fucked around with this message at 18:02 on Jun 19, 2016 |
|
#
?
Jun 19, 2016 18:00
|
|
|
:edit wrong thread:
|
|
#
?
Jun 19, 2016 19:26
|
|
|
Palladium posted:Hard to believe used AMD 1100Ts are still moving for ~$180. Is there a faster AMD processor yet? Or will it be unbeaten until zen?
|
|
#
?
Jun 20, 2016 17:57
|
|
|
An FX-8320E is faster but not appreciably so, so in practical terms it's unbeaten as the AMD processor to have until Zen.
|
|
#
?
Jun 20, 2016 18:59
|
|
|
a 6~+ year old CPU is the best AMD has currently? Jeez.
|
|
#
?
Jun 20, 2016 21:35
|
|
|
EdEddnEddy posted:a 6~+ year old CPU is the best AMD has currently? And now you know why everyone's beating it over zen in the hopes that AMD regains performance cpu viability.
|
|
#
?
Jun 20, 2016 22:07
|
|
|
NihilismNow posted:Is there a faster AMD processor yet? Or will it be unbeaten until zen? If we're not joking around, the 9590 is the fastest. It's just VRM cookingly power hungry.
|
|
#
?
Jun 20, 2016 22:38
|
|
|
HalloKitty posted:If we're not joking around, the 9590 is the fastest. It's just VRM cookingly power hungry. Holy poo poo! "32nm Vishera 220W" I had no idea AMD had a processor with that much TDP. loving insane.
|
|
#
?
Jun 20, 2016 22:44
|
|
redeyes posted:Holy poo poo! "32nm Vishera 220W" Yeah, it's pretty hilarious, a fair number of mobos with the right socket are not actually compatible with the thing because it will quite literally burn out(as in smoke and in rare cases actual fires) the VRMs because they are not built to handle that sort of power, it should have come with a fire hazard warning on the box. 
|
|
|
#
?
Jun 20, 2016 22:59
|
|
|
Bulldozer: A bad processor? A shocking inside report, coming up at 11, only on The Circa 2012 News Network
|
|
#
?
Jun 21, 2016 01:08
|
|
|
AVeryLargeRadish posted:Yeah, it's pretty hilarious, a fair number of mobos with the right socket are not actually compatible with the thing because it will quite literally burn out(as in smoke and in rare cases actual fires) the VRMs because they are not built to handle that sort of power, it should have come with a fire hazard warning on the box. For comparison, that's about as much as an overclocked 5820K pulls at 4.3 GHz while running Prime95 SmallFFT, and that's considered a notoriously power-hungry processor for a consumer unit and an incredibly power-hungry test. I looked at it back in 2014 when I was rebuilding. The worst thing was that all the motherboards with the right chipset to handle that fuckoff-huge amount of power were like a billionty dollars, meaning for the same price you could just get an Intel processor that didn't turn your PC case into an EZ Bake Oven. I had AMD since I was a kid, I want to like them, but it's just impossible nowadays, so I rebuilt with a 4690K instead. I like AM1 for home server stuff and FM2 is nice for tiny momputers, other than that their stuff is just decimated at every step by Intel and maybe a 750 Ti if you need. Bulldozer is actually OK if you keep it near its stock clocks and do something that isn't latency-sensitive and exposes enough threads to fully occupy it. So like video encoding and stuff. But otherwise it's literally a step backwards from Phenom II, which is why prices are still nuts for that chip. I had an X4 830, it was pretty great for most stuff, especially considering I paid almost nothing for it. The 4690K I replaced it with curbstomped it for gaming though. Paul MaudDib fucked around with this message at 01:27 on Jun 21, 2016 |
|
#
?
Jun 21, 2016 01:14
|
|
|
EdEddnEddy posted:a 6~+ year old CPU is the best AMD has currently? Like the 2600k 
|
|
#
?
Jun 21, 2016 01:32
|
|
|
BIG HEADLINE posted:Of course, in the very same article he claims that "SATA drives will become passe" and that the ~futcha~ is in *PCIe* drives. Granted, that statement could include M.2 drives connected to the PCIe interface, but still... Yeah, I'm guessing that's what he meant. NVMe has finally brought PCIe SSDs to the unwashed masses - either via M.2 or by PCIe adapter sleds. And it's wicked fast. Now if they could just fix boot times from NVMe, loading option-roms takes too much time. NihilismNow posted:The answer is clearly 6800k. 6800K is $400 at MicroCenter versus $320 for the 5820K or $290 for the 6700K. X99 vs Z170 motherboards are pretty much a wash either way. I'm more than willing to listen if there's a case to be made for Broadwell-E but everyone seemed underwhelmed with it given the $80 price bump. Whoopee, it draws 5 watts less than Haswell-E, I'd still have to upgrade my cooling  8-thread vs 12-thread is the debate I'm stuck on. I'm pretty sold on 12-thread where I think DX12/Vulkan games are headed and encoding tasks I do, but the heat and potential loss of single-core performance still worry me. I'll probably just install a 280mm AIO and suck it up on the single-threaded performance. My 4690K is now in my mITX Vive PC and my desktop is running a 4170 that was formerly in my TV PC and it sucks, it's bottlenecking my new 980 Ti like crazy and it sucks rear end at encoding. I gotta do something real soon here. Paul MaudDib fucked around with this message at 01:40 on Jun 21, 2016 |
|
#
?
Jun 21, 2016 01:35
|
|
|
redeyes posted:Holy poo poo! "32nm Vishera 220W" For a  comparison, an i5 6500 only draws 38W more power from idle to full CPU load, for the entire system while having ~50% more ST performance.AVeryLargeRadish posted:Yeah, it's pretty hilarious, a fair number of mobos with the right socket are not actually compatible with the thing because it will quite literally burn out(as in smoke and in rare cases actual fires) the VRMs because they are not built to handle that sort of power, it should have come with a fire hazard warning on the box. By the time one factors in the cost of an AMD FX mobo that doesn't blow itself up when OCed, a beefier PSU and aftermarket HSF, and increased power draw for the privilege of being stuck with an antiquated chipset, the equivalent Intel rig might actually end up being cheaper in the long run. "But-but-but Intel fanbois dun know how they are getting ripped off for not buying cheaper AMD lulz." Palladium fucked around with this message at 05:32 on Jun 21, 2016 |
|
#
?
Jun 21, 2016 05:12
|
|
|
|
| # ? May 11, 2024 08:08 |
|
|
Even makes me feel less bad about the power usage of my 3930K Sandybridge-E. Really it idles rather cool, but under full load the system can eat some watts. Good thing we got Solar earlier this year and so far it has been canceling out pretty much all usage.  Might look to upgrade when whatever after X99 arrives but until then, at 4.6Ghz this chip eats everything I can throw at it. Just needs a 1080Ti for that VR Streaming.
|
|
#
?
Jun 21, 2016 16:42
|
|