|
Also the TIM thing is a stealth way of forcing you to void your warranty and also saves intel a few bucks since soldering dies is quite complex
|
#
?
Jun 4, 2017 01:58
|
|
|
|
| # ? Apr 27, 2024 22:24 |
|
|
Malcolm XML posted:Also the TIM thing is a stealth way of forcing you to void your warranty and also saves intel a few bucks since soldering dies is quite complex It's still inexcusable on a $600+ part. Unless coffee lake is amazeballs I'm going with the next version of Zen.
|
|
#
?
Jun 4, 2017 04:49
|
|
|
sincx posted:And apparent RAID 1/10 is $99 and RAID 5 is $199 or $299. And if there's any non-Intel NVME drive in the array it doesn't allow you to boot off the array.  Also none of the HEDT CPUs support QuickSync because that would be handled by the non-existant iGPU part. Not that you'd want to use QuickSync with 8+ cores available but still interesting to know. eames fucked around with this message at 07:13 on Jun 4, 2017 |
|
#
?
Jun 4, 2017 07:11
|
|
|
How hard it can be to use good thermal paste? I've used the same 3 gram tube of Arctic Silver 3 since the dawn of time, and temps are the same after all the years. Arctic Silver 5 is under $10 and you can use it for many, many CPU's. So what is the issue with the lovely pastes?!
|
|
#
?
Jun 4, 2017 08:43
|
|
|
"lovely paste" has upsides for Intel. Saves a bit of money, introduces planned obsolescence through TIM degradation and forces overclockers to void warranties. Win-win-win for Intel.
|
|
#
?
Jun 4, 2017 09:04
|
|
|
I can't believe Raid Keys are going to be a thing for the X299. I understand if raid keys were a thing for an enterprise demographic, but for consumers/prosumers?
|
|
#
?
Jun 4, 2017 18:46
|
|
|
At this point, I'll probably just get a Threadripper out of principle, even though the lower IPC bugs me. That bitching video from Linus Tech Tips pretty much condenses all issues with Intel for this release.
|
|
#
?
Jun 4, 2017 18:56
|
|
|
There is practically no reason to use Intel raid on a consumer (or enterprise) board. NVMe is so fast there is not any benefit to RAID other than headaches. I've had trouble with Intel RAID on servers too, so I don't use it unless I have to.
|
|
#
?
Jun 4, 2017 19:01
|
|
|
Did Intel just ignore soft raid and zfs or something Who runs raid on a desktop anyway?
|
|
#
?
Jun 4, 2017 19:09
|
|
|
If you haven't figured it out, it's mostly a bunch of AMD people trumpeting a perceived disadvantage of Intel to push AMD products as an alternative. Much like the whole "Intel hates overclockers!" thing a few weeks ago. Nobody actually uses Intel Raid on a desktop system, there is no reason a consumer needs dozens of terabytes of direct-attached bootable RAID storage and if you do then ponying up $100 for the RAID key is not going to be a big deal. Just use LVM or ZFS like a regular autist. The TIM is a much bigger deal actually.
|
|
#
?
Jun 4, 2017 19:43
|
|
|
Paul MaudDib posted:If you haven't figured it out, it's mostly a bunch of AMD people trumpeting a perceived disadvantage of Intel to push AMD products as an alternative. Much like the whole "Intel hates overclockers!" thing a few weeks ago. People want even more speed in their raid nvme drives we must reach near ddr speeds in our hardrives.
|
|
#
?
Jun 4, 2017 19:54
|
|
|
The USP of Intel VROC is that it bypasses the chipset, goes directly into the CPU and is bootable. Asus demoed 13 Gigabytes per second with eight drives but the theoretical maximum is 128 Gigabyte per second (PCIe 3.0 x16). Anything running off the chipset would be bottlenecked by DMI (PCIe 3.0 x4). As for who would need that much bandwidth on a platform that appears to be marketed strictly to gamers, no idea. Gotta keep those loading boot and loading screens short.  e: actually Linus is the kind of guy that would run a production server off eight SSDs in RAID 0, perhaps that's why he made the video eames fucked around with this message at 20:07 on Jun 4, 2017 |
|
#
?
Jun 4, 2017 20:04
|
|
|
So basically people that don't think they need a professional setup and actually should but think "that's for corporate techies, not me." Sounds like a lot of the market given they probably have more money than sense and constantly are upgrading stuff and acting as beta testers effectively.
|
|
#
?
Jun 4, 2017 20:10
|
|
|
eames posted:The USP of Intel VROC is that it bypasses the chipset, goes directly into the CPU and is bootable. I mean, apart from the "Intel RAID" part, what would stop you from putting 4 NVMe cards on a sled and just using it as 4 NVMe drives that then use soft-RAID or ZFS? It wouldn't be bootable, but it should come up once you're into the OS, right?
|
|
#
?
Jun 4, 2017 20:17
|
|
|
Paul MaudDib posted:If you haven't figured it out, it's mostly a bunch of AMD people trumpeting a perceived disadvantage of Intel to push AMD products as an alternative. Much like the whole "Intel hates overclockers!" thing a few weeks ago. Its a user hostile development that's also completely useless to gamers and most workstation users Par for the modern Intel course
|
|
#
?
Jun 4, 2017 20:23
|
|
|
It's for taking that $100 per workstation perc hw raid controller out of dell's pocket when an enterprise desktop it manager buys five hundred new overspecced desktops.
|
|
#
?
Jun 4, 2017 20:23
|
|
|
eames posted:As for who would need that much bandwidth on a platform that appears to be marketed strictly to gamers, no idea. Gotta keep those loading boot and loading screens short.
|
|
#
?
Jun 4, 2017 20:53
|
|
|
Paul MaudDib posted:I mean, apart from the "Intel RAID" part, what would stop you from putting 4 NVMe cards on a sled and just using it as 4 NVMe drives that then use soft-RAID or ZFS? It wouldn't be bootable, but it should come up once you're into the OS, right? Nothing and yes unless they wilfully implemented a regression, that's possible with any system that supports PCIe Gen 3.
|
|
#
?
Jun 4, 2017 20:55
|
|
|
PCjr sidecar posted:It's for taking that $100 per workstation perc hw raid controller out of dell's pocket when an enterprise desktop it manager buys five hundred new overspecced desktops. Who buys raid in a workstation these days? Most purchasers are buying SSDs and those are still too expensive to raid E: with a storage hdd i guess
|
|
#
?
Jun 4, 2017 21:02
|
|
|
There is also the intel Volume Management Device (VMD) stuff which supports setting up volumes using pcie in the bios, not sure if they support various RAID flavours or if there is some overlap with this VROC stuff. It's more of an enterprise thing too I think.
|
|
#
?
Jun 4, 2017 21:04
|
|
|
I honestly don't understand why Intel doesn't sell an "XE" chip with a bare die, save for the fact that every cooler maker might have to retool their AIOs or HSFs (or mail/sell adapters) to compensate for the loss of the thickness of the IHS.
|
|
#
?
Jun 5, 2017 06:27
|
|
BIG HEADLINE posted:I honestly don't understand why Intel doesn't sell an "XE" chip with a bare die, save for the fact that every cooler maker might have to retool their AIOs or HSFs (or mail/sell adapters) to compensate for the loss of the thickness of the IHS. Bare dies are more delicate, GPU makers can get away with it because there is a cooler on top and they can hold the consumer responsible if the chip gets damaged, CPU makers not so much.
|
|
|
#
?
Jun 5, 2017 06:34
|
|
|
How delicate things are depends on the cooler. Air coolers for OC are pretty heavy and towering, they create interesting forces on the die, given that it's quasi levering from the top two screws. Watercooling should be more on the foolproof side.
|
|
#
?
Jun 5, 2017 13:27
|
|
|
AVeryLargeRadish posted:Bare dies are more delicate,
|
|
#
?
Jun 5, 2017 14:40
|
|
PC LOAD LETTER posted:Ship it with a shim glued in place or something. A few idiots will still manage to crack the die but by and large it'd probably work just fine I bet. I think the rate of breakage would be higher than you think, and when they break Intel has to replace them because they shipped them that way, it's just a headache they don't want to deal with. If you want a delided CPU do it yourself, that is part of the risk inherent in running hardware outside of spec.
|
|
|
#
?
Jun 5, 2017 15:09
|
|
|
Didn't AMD ship bare CPUs back in the day? Plus, being Intel, I'm sure they'd manage to throw some legalese on there where they are not responsible for any physical damage to the chip, they recommend it be installed by a certified professional, etc. I agree it'd still be a bit of a headache, but with a decent shim setup it shouldn't be too bad. That said, the fact remains that Intel simply doesn't care enough to bother catering to what would be a pretty small demographic.
|
|
#
?
Jun 5, 2017 15:16
|
|
|
I read that as "AMD ship bare back CPUs". That's one hot processor.
|
|
#
?
Jun 5, 2017 15:30
|
|
|
DrDork posted:Didn't AMD ship bare CPUs back in the day? Yeah, and it was easy to break them. They started including little rubber pads in the corners in an attempt to make sure you put the HSF on level.
|
|
#
?
Jun 5, 2017 15:43
|
|
|
redeyes posted:I read that as "AMD ship bare back CPUs". That's one hot processor. POS my chip hole
|
|
#
?
Jun 5, 2017 15:44
|
|
|
Boiled Water posted:POS my chip hole Hot chip mounting POV video.mpg
|
|
#
?
Jun 5, 2017 16:28
|
|
|
These days delidding is a pretty simple and foolproof procedure anyway as long as you get a proper kit made for your particular CPU, and don't try to eyeball it with a razerblade. Of course, your warranty is toast, but if you know what you're doing with the repaste (like if you want to use LM you need to tape off the PCB etc.) then it's not really a massive deal. Honestly, the biggest problem of it for me is 
|
|
#
?
Jun 5, 2017 16:49
|
|
|
HalloKitty posted:Yeah, and it was easy to break them. They started including little rubber pads in the corners in an attempt to make sure you put the HSF on level. Even if they're not going to cover it under warranty though, I can see why intel would rather just avoid opening that can of worms. Surely some people would try and make them out to be the bad guy for not covering it when people inevitably break them.
|
|
#
?
Jun 5, 2017 17:17
|
|
Col.Kiwi posted:This is true. I don't think they covered it under warranty if you broke it though? Pretty sure it was just considered user error. Hardware warranties never cover customer induced physical damage. Like if you bend a pin in your LGA socket and try to get warranty, board manufacturers are just going to offer to repair it for a fee. It may be a fragile thing but still the bottom line is you broke it. IIRC, Intel's warranty used to cover just about anything including user damage, but they changed it a few years back when delidding started becoming really popular.
|
|
|
#
?
Jun 5, 2017 17:23
|
|
|
Shrimp or Shrimps posted:Honestly, the biggest problem of it for me is
|
|
#
?
Jun 5, 2017 17:57
|
|
|
Combat Pretzel posted:For me it's the effort required in future. Like watching temps and then eventually repasting things. At least that was implied necessary at some point. If you use good paste it'll last pretty much forever. And there've been more than one case of lovely factory TIM basically failing after a few years, requiring a repaste anyhow. Soldered chips have largely been immune from this issue, but it seems reasonable to assume that any Intel chip with stock TIM will need to be "fixed" at some point in the future.
|
|
#
?
Jun 5, 2017 18:43
|
|
|
.
sincx fucked around with this message at 05:55 on Mar 23, 2021 |
|
#
?
Jun 5, 2017 22:48
|
|
|
Platfrom
|
|
#
?
Jun 5, 2017 22:52
|
|
|
TIM unlock $99
|
|
#
?
Jun 5, 2017 22:53
|
|
|
Wasn't the real problem the thickness of the glue they used causing the heat spreader to be lifted from the die a tiny amount, and that the actual TIM was fine?
|
|
#
?
Jun 5, 2017 23:40
|
|
|
|
| # ? Apr 27, 2024 22:24 |
|
|
priznat posted:TIM unlock $99 Silicon Lottery charges $49, plus shipping.
|
|
#
?
Jun 5, 2017 23:41
|
|