|
SourKraut posted:Thunderbolt 3 PCIe cards are available for $50? Well, $60, but yes.
|
# ? Sep 11, 2017 17:29 |
|
|
# ? May 10, 2024 19:35 |
|
DrDork posted:Well, $60, but yes. That requires a thunderbolt header and only works with specific expensive ASUS boards. It's not a general solution.
|
# ? Sep 11, 2017 17:50 |
|
As far as I know there's no drop-in You Have Thunderbolt Now card yet. Grain of salt or six, but if these are true I'm somewhere between disappointed and anxious. Disappointed because I was sort of holding out for "7700K, with another half a die", but anxious because I want to see what happens to the market. I'm hoping to pick up a used 7700K at a decent price, basically.
|
# ? Sep 11, 2017 18:02 |
|
That's way lower than the previously leaked benchmarks. Guess we will have to wait and see.
|
# ? Sep 11, 2017 18:28 |
|
I'm taking this with the highest peak of salt imaginable. What the hell happened to Intel's IPC advantage in those benchmarks? Edit: and if those benchmarks are accurate, all the more reason to hold onto my oc'd 2500K rig until it goes kaput.
|
# ? Sep 11, 2017 19:18 |
|
SourKraut posted:Thunderbolt 3 PCIe cards are available for $50? You cant just plug a thunderbolt card into a motherboard. The MB has to support it.
|
# ? Sep 11, 2017 19:19 |
|
I suspect the undelidded throttling problem already on the 7700k is even worse on the 8700k hence the terrible bench
|
# ? Sep 11, 2017 20:01 |
|
BurritoJustice posted:That requires a thunderbolt header and only works with specific expensive ASUS boards. It's not a general solution. I do happen to have both a asus tb3 card and an inexpensive motherboard(z170a). The same card is supported on the z270a prime which is the cheapest or so option for their z270 catalog. I expect the upcoming z370a prime to do the same.
|
# ? Sep 11, 2017 20:07 |
|
Is Z370 going to increase the size of the DMA bus to the chipset, or is it still "24 lanes" all going through a single 3.0x4 link?
|
# ? Sep 16, 2017 13:54 |
|
Paul MaudDib posted:Is Z370 going to increase the size of the DMA bus to the chipset, or is it still "24 lanes" all going through a single 3.0x4 link? Last I was aware, the Z370 is basically a rebranded Z270 with some very minor upgrades, so I don't think anything as substantial as the DMI link is changing.
|
# ? Sep 16, 2017 14:04 |
|
I very much doubt that we'll ever see more than an x4 link between chipset and CPU. There's very little reason to want it, except for wanting to plug in higher bandwidth devices, which generally just call for CPU lanes. In my ideal world, Icelake will come out with PCIe 4 (+ DMI 4, improving bandwidth to chipset) and 20 CPU lanes on the consumer boards (GPU + NVMe), and I'll squeal with delight. Who the hell knows what will happen, though, everybody's mum about a time frame for improving this stuff. I'm not sure if I'm a weirdo for wanting improved IO, or if they're just blind to the demand for it. I guess NVMe has kinda splashed onto the stage pretty quick, but hell, AMD saw the demand for 20 CPU lanes because of it already, right?
|
# ? Sep 16, 2017 17:59 |
|
I want improved Io because then manufacturers will create storage devices to use it. I know, small files and optane etc.
|
# ? Sep 16, 2017 18:10 |
|
With multi-gpu dead and NVME raid still not a thing (not that its in the realm of reason anyway with current prices), I really dunno what i would use more I/O for. As it stands, the only chipset PCIe lanes I'm using or plan to use are 4x for an NVME drive and the integrated USB 3.1 Gen 2 on the motherboard that I think is hanging off a 1x. In a workstation world, maybe you would want multi gpus for deep learning or higher end NICs (infiband, 40gbe etc) or somethin, but on the consumer side there isn't much use for a bunch of lanes. Cygni fucked around with this message at 18:32 on Sep 16, 2017 |
# ? Sep 16, 2017 18:30 |
|
VCZ: Intel Z390 to support 8C/16T CPUs in H2 2018
|
# ? Sep 16, 2017 20:26 |
Considering even an arm soc can have 4k die pads, I don't think intel has much room to add more lanes without increasing the die size, especially considering an x86 processor sucks up magnitudes more power.
|
|
# ? Sep 16, 2017 20:53 |
|
So basically Intel Threadripper?
|
# ? Sep 17, 2017 00:22 |
|
blowfish posted:So basically Intel Threadripper? maybe, ifs it going up against the 1900.
|
# ? Sep 17, 2017 00:36 |
|
lmao if they paste it or it isn't just a xeon e3
|
# ? Sep 17, 2017 02:04 |
|
They're already pasting the 18 core models, 8 cores is no question for pasting.
|
# ? Sep 17, 2017 02:12 |
|
God drat i really wish they would stop loving pasteing
|
# ? Sep 17, 2017 04:55 |
|
Scarecow posted:God drat i really wish they would stop loving pasteing But what about the $0.50/unit cost
|
# ? Sep 17, 2017 10:29 |
|
Think of the indium
|
# ? Sep 18, 2017 19:15 |
|
When's the expected release of all the new boards and desktop Coffee Lake?
|
# ? Sep 20, 2017 04:30 |
|
Tab8715 posted:When's the expected release of all the new boards and desktop Coffee Lake? Oct 5th.
|
# ? Sep 20, 2017 04:32 |
|
Don Lapre posted:Oct 5th. drat. New PC next month!
|
# ? Sep 20, 2017 04:52 |
|
Tab8715 posted:drat. New PC next month! Just be aware that a few hundred thousand to a million-plus people are thinking this same thing - you might want to be on the lookout for pre-order bundles of CPU+Motherboard.
|
# ? Sep 20, 2017 06:05 |
|
Well, I’ll do what I’m able. Any preference on motherboards?
|
# ? Sep 20, 2017 07:42 |
|
Tab8715 posted:Well, I’ll do what I’m able. Any preference on motherboards?
|
# ? Sep 20, 2017 11:49 |
|
Intel reportedly pushing Cannon Lake to end of 2018 Intel's NUC roadmap leaked, it includes a pair of Skull Canyon refreshes with discrete graphics codenamed "Hades Canyon" and "Hades Canyon VR", the latter with a 4C8T K-series desktop processor at 100W.
|
# ? Sep 20, 2017 19:32 |
|
Paul MaudDib posted:Intel reportedly pushing Cannon Lake to end of 2018 I wonder how much of this is "10nm is hard" and how much is, "Oh poo poo, we actually have to do something about Threadripper."
|
# ? Sep 20, 2017 19:36 |
|
I’d like to see the new Gen5 GPU and at least start to replace the 1080p standard. Modern game titles would also for the first time become playable on laptops without an extra GPU. Plus, a single digit nanometer processor comes with bragging rights.
|
# ? Sep 20, 2017 21:35 |
|
Gyrotica posted:I wonder how much of this is "10nm is hard" and how much is, "Oh poo poo, we actually have to do something about Threadripper." threadripper? probably not. raven ridge yes. epyc double yes
|
# ? Sep 20, 2017 22:52 |
Gyrotica posted:I wonder how much of this is "10nm is hard" and how much is, "Oh poo poo, we actually have to do something about Threadripper." Give me '10nm is hard'!!! also: i wonder if that dgpu nuc is the rumored licensing by intel of radeon stuff? Watermelon Daiquiri fucked around with this message at 01:12 on Sep 21, 2017 |
|
# ? Sep 21, 2017 00:58 |
|
Watermelon Daiquiri posted:Give me '10nm is hard'!!! 10nm is a rat bastard because you need EUV to get the critical logic layers and parts of the cache down in size and therefore down in cost. The fab manufacturing companies just released a commercially viable litho system for 10nm. Also keep in mind that a modern chip with the dual or quad multipatterning layers takes between 4 and 6 months from the day the wafers show up to the day the finished chips go in a bin.
|
# ? Sep 21, 2017 01:16 |
Methylethylaldehyde posted:10nm is a rat bastard because you need EUV to get the critical logic layers and parts of the cache down in size and therefore down in cost. The fab manufacturing companies just released a commercially viable litho system for 10nm. quad i wouldn't doubt, but dual patterned stuff like 14nm can be done within 70 days if they scoot their asses lol However for engineering wafers, they frequently do tests of each sub process alone so they don't have to wait months to find out that the mandrell mask has a defect lol
|
|
# ? Sep 21, 2017 01:39 |
|
Black Hat '17: How to Hack a Turned-Off Computer, or Running Unsigned Code in Intel Management Engine yikes
|
# ? Sep 21, 2017 01:40 |
|
repiv posted:Black Hat '17: How to Hack a Turned-Off Computer, or Running Unsigned Code in Intel Management Engine What I want to know is whether this is a legitimate "oops" vulnerability or something Intel put in both officially and intentionally for the NSA and GCHQ to exploit, and it's just a matter of time before such things are found by the public, or it's being revealed piecemeal from sources like Snowden's trove, etc. BIG HEADLINE fucked around with this message at 03:46 on Sep 21, 2017 |
# ? Sep 21, 2017 02:32 |
|
Watermelon Daiquiri posted:Give me '10nm is hard'!!! That's what she said.
|
# ? Sep 21, 2017 03:39 |
|
Presumably this is an escalation of the exploit reported earlier this year?
|
# ? Sep 21, 2017 08:07 |
|
|
# ? May 10, 2024 19:35 |
|
BIG HEADLINE posted:What I want to know is whether this is a legitimate "oops" vulnerability or something Intel put in both officially and intentionally for the NSA and GCHQ to exploit, and it's just a matter of time before such things are found by the public, or it's being revealed piecemeal from sources like Snowden's trove, etc.
|
# ? Sep 21, 2017 08:37 |