Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us $3,400 per month for bandwidth bills alone, and since we don't believe in shoving popup ads to our registered users, we try to make the money back through forum registrations.
«562 »
  • Post
  • Reply
Shaocaholica
Oct 29, 2002

Fig. 5E


And the ~2008+ Dell bios had mouse support and more UI stuff. Way before UEFI Winamp junk. Has anyone made a UEFI bios thatís functionally clean? Big OEMs donít count. I mean in the consumer space.

Adbot
ADBOT LOVES YOU

D. Ebdrup
Mar 13, 2009



Shaocaholica posted:

And the ~2008+ Dell bios had mouse support and more UI stuff. Way before UEFI Winamp junk. Has anyone made a UEFI bios thatís functionally clean? Big OEMs donít count. I mean in the consumer space.
For all of my dislike of the firmware of SuperMicro "gamer" boards, the actual options that are in the various sections match the server boards layout to a tee; it's just presented with a bit of fluff around it, and even then it's quite muted compared to other more garish designs seen on RGB flavoured boards.
Anandtech has a gallary for the previous generation, and I would be surprised if it's changed.

HalloKitty
Sep 30, 2005

Adjust the bass and let the Alpine blast


Shaocaholica posted:

And the ~2008+ Dell bios had mouse support and more UI stuff. Way before UEFI Winamp junk. Has anyone made a UEFI bios thatís functionally clean? Big OEMs donít count. I mean in the consumer space.

The msi non-max boards with the zen 2 update have a clean old school look, out of necessity

Cygni
Nov 12, 2005

raring to post



https://twitter.com/momomo_us/statu...829078007242753

So Asrock (some), MSI (most), and Giga (all) saying their boards will work with PCIe 4.0... and Asus aint sayin poo poo. Honestly after the AMD compatibility drama catching some of these board partners out last dang week, I kinda think Asus being conservative is the right position for an OEM to take.

orcane
Jun 13, 2012



Fun Shoe

Doesn't that require the next CPU generation anyway? You'd think they'll want to sell Z590 boards by then

JawnV6
Jul 4, 2004

So hot ...

how would per-core multipliers even work, you'd need BGF's all over the place

D. Ebdrup
Mar 13, 2009



JawnV6 posted:

how would per-core multipliers even work
Very carefully.

Cygni
Nov 12, 2005

raring to post



JawnV6 posted:

how would per-core multipliers even work, you'd need BGF's all over the place

yeah that setting in current day OC BIOS reflects turbo frequency, not actual per core multiplers with different clock domains.

That said, I would be interested to see how Winders handles the Snapdragon 835, since it has separate clock domains for the bigLITTLE... might be a good hint as to how they plan to handle Alder Lake and Lakefield's multiple clock domains.

Paul MaudDib
May 2, 2006

"Tell me of your home world, Usul"


EVGA has B-stock X299 Dark for $170 free shipping. No idea what I'd use it for, but drat if that isn't a killer price on one of the best ricer X299 boards.

I suppose nobody else has any idea either, that's why they're on super-special.

JawnV6
Jul 4, 2004

So hot ...

Cygni posted:

yeah that setting in current day OC BIOS reflects turbo frequency, not actual per core multiplers with different clock domains.
ohh, so 1 core we've got headroom 52x, 4 cores we're limited 48x. gotcha, makes a lot more sense. thanks!

Cygni posted:

That said, I would be interested to see how Winders handles the Snapdragon 835, since it has separate clock domains for the bigLITTLE... might be a good hint as to how they plan to handle Alder Lake and Lakefield's multiple clock domains.
back when i was more involved with this stuff a big idea was hiding all complexity behind ACPI. like it's not some extra core, it's just a power state on "core" 0! it "works" kinda sorta but doesn't take full advantage of the HW (obv. can't have something running on both of them), then you've also got to worry about ISA's being an exact match or some convoluted scheme to trap on a AVX #UNDEF on a LITTLE core and have machinery to punt it over to the big. it sounded like the same kind of no-go zone as core hopping.

D. Ebdrup
Mar 13, 2009



Intels new Xeon W-1200, specifically the W-1290, is exactly what I've been looking for for my new workstation.
The new 400 series chipset also supports vPro across the whole line.

Shaocaholica
Oct 29, 2002

Fig. 5E


D. Ebdrup posted:

Intels new Xeon W-1200, specifically the W-1290, is exactly what I've been looking for for my new workstation.
The new 400 series chipset also supports vPro across the whole line.

For work or personal?

D. Ebdrup
Mar 13, 2009



Shaocaholica posted:

For work or personal?
Personal.
I'm looking to build a system that's got some remote capabilities, ECC, and as much memory as possible.
It'll be running FreeBSD and doing routing+ipfw for my FTTH, as well as hosting NFS+krb5i filesharing on top of ZFS for both LAN and WAN, daily driver desktop OS with Firefox as a browser, and using bhyve as a hypervisor for Windows 10 for gaming.
Since I'll be doing virtualization, I also need the firmware and the onboard NICs to support SR-IOV, otherwise I'm gonna need even more pci-ex lanes.

D. Ebdrup fucked around with this message at 19:06 on May 14, 2020

SamDabbers
May 26, 2003



Fallen Rib

I thought about doing something similar, but decided against putting the routing on my workstation so that Internet stays up even when I'm updating/rebooting the box in case things go sideways and I have to use my laptop to look up docs, download a patch, etc.

How's bhyve work for GPU passthrough these days? Can it mask itself sufficiently to work with NVIDIA drivers and their consumer GPUs?

D. Ebdrup
Mar 13, 2009



SamDabbers posted:

I thought about doing something similar, but decided against putting the routing on my workstation so that Internet stays up even when I'm updating/rebooting the box in case things go sideways and I have to use my laptop to look up docs, download a patch, etc.

How's bhyve work for GPU passthrough these days? Can it mask itself sufficiently to work with NVIDIA drivers and their consumer GPUs?
My laptop has a HSPA modem, and I've got a cheap LTE-A modem on the way (I succeeded in finding a mini-pcie-ex modem with LTE! That in itself is a miracle!). Both support USB CDC ECM/NCM, so they just appear to FreeBSD as a cdce(4)-based NIC that you request a DHCP address from, instead of having to do PPP.

As for reboots, FreeBSD gets security updates every 10-50 days, but not all of them include the kernel, so since FreeBSD is smart about patch levels, reboots are only needed if the kernel is updated.
Unless it's something like the latest libalias (read: ipfw in-kernel NAT, that I use) or cryptodev security advisories (both of which were quite impressive!), I can usually justify putting off a reboot until I go to bed, when I won't be using it.
It's not like there's any risk of rebooting when I have ZFS and boot environments, as if it fails to reboot, it'll just boot up in the old boot environment and I can look at what went wrong in the morning.

Nope, no proper GPU passthrough. Mostly because nvidia are still being complete shitheels about complete IOMMU-based device pass-through (although AMD aren't good about SR-IOV pass-through either).
Which is to say, you can pass through the GPU to a FreeBSD guest and have the FreeBSD driver attach to it, but if you try it with Windows, it just shrugs at you and acts like a mopey teenager.
I'll likely end up with an AMD graphics card, but I spent most of my mopey teenage years with an Intel CPU and AMD graphics card, so it'll be familiar.

D. Ebdrup fucked around with this message at 20:27 on May 14, 2020

Hed
Mar 31, 2004



Fun Shoe

D. Ebdrup posted:

My laptop has a HSPA modem, and I've got a cheap LTE-A modem on the way (I succeeded in finding a mini-pcie-ex modem with LTE! That in itself is a miracle!). Both support USB CDC ECM/NCM, so they just appear to FreeBSD as a cdce(4)-based NIC that you request a DHCP address from, instead of having to do PPP.

I'd love to hear more about the LTE-A modem--like which one you used. Especially interested in one for FreeBSD.

D. Ebdrup
Mar 13, 2009



Hed posted:

I'd love to hear more about the LTE-A modem--like which one you used. Especially interested in one for FreeBSD.
Turns out, I forgot LTE nomenclature and misremembered the 4 in cat5 as an A. orz
It's a Huawei ME909s-120 which is supported in every supported branch. It does LTE cat4, so it's ~150/50Mbps depending on signal and how much the shared medium is being utilized.
At least the latency should better, as that's one of the areas that LTE made decent improvements on over HSPA, if memory serves?

Malcolm XML
Aug 8, 2009

I always knew it would end like this.


JawnV6 posted:

ohh, so 1 core we've got headroom 52x, 4 cores we're limited 48x. gotcha, makes a lot more sense. thanks!

back when i was more involved with this stuff a big idea was hiding all complexity behind ACPI. like it's not some extra core, it's just a power state on "core" 0! it "works" kinda sorta but doesn't take full advantage of the HW (obv. can't have something running on both of them), then you've also got to worry about ISA's being an exact match or some convoluted scheme to trap on a AVX #UNDEF on a LITTLE core and have machinery to punt it over to the big. it sounded like the same kind of no-go zone as core hopping.

Unless you're samsung and dgaf https://link.medium.com/arrb9RG8u6

WhyteRyce
Dec 30, 2001



If you had to choose between an i7 7700k or i5 8500 hand-me downs for gaming, it looks like the 7700k is the better choice?

wargames
Mar 16, 2008

official yospos cat censor


WhyteRyce posted:

If you had to choose between an i7 7700k or i5 8500 hand-me downs for gaming, it looks like the 7700k is the better choice?

I think the 7700l would be better.

eames
May 9, 2009



Computerbase benchmarked Coffee Lake vs Ryzen with various memory speeds and two different GPUs.
The Intel chips still like fast, optimized RAM, particuarly when overclocked. Most sites only publish stock vs stock results, but the overclocked comparison shows a sizeable gap in both FPS and frametimes.

here's the link: https://www.computerbase.de/2020-05...-9-3900x-ram-oc

e: attachment is way too large but the post form doesn't give me an option to modify it.

Only registered members can see post attachments!

eames fucked around with this message at 09:27 on May 19, 2020

mobby_6kl
Aug 9, 2009

"You are the best poster... do not let anyone say otherwise."


Comet Lake reviews are out. Only had a couple of minutes to look through but Tl;dr seems to be that they're pretty fast but at the expense of horrific power consumption.

https://www.anandtech.com/show/1578...ake-we-go-again

Some Goon
Jan 6, 2013

A golden helix streaked skyward from the Helvault. A thunderous explosion shattered the silver monolith and Avacyn emerged, free from her prison at last.



Yeah, turns out the spec sheets told you all you needed to know, its a faster, and thus hotter, Coffee Lake with more cores.

HalloKitty
Sep 30, 2005

Adjust the bass and let the Alpine blast


Some Goon posted:

Yeah, turns out the spec sheets told you all you needed to know, its a faster, and thus hotter, Coffee Lake Skylake with more cores.

It's OK, but I'm a bit surprised that 10 cores at 4.9GHz burning an enormous amount of power couldn't catch up to a 3900X in productivity tasks. I thought maybe the sheer clock advantage would have helped there.
It really is getting a bit desperate now though, they really need a process shrink or a newer architecture, we're reaching Peak Skylakeô

HalloKitty fucked around with this message at 16:52 on May 20, 2020

ratbert90
Feb 12, 2009
JUST FUCKING STOP, JESUS H. CHRIST


So nothing has changed? AMD is still better for everything but very specific single-threaded productivity tasks and AMD has a 5%~ FPS drop for games.

AMD still runs cooler, is cheaper, and draws less power as well. Am I missing something?

an actual dog
Nov 18, 2014

woof woof
I am a dog and I'm posting online. it's good to post online with friends.
woof bark woof



ratbert90 posted:

So nothing has changed? AMD is still better for everything but very specific single-threaded productivity tasks and AMD has a 5%~ FPS drop for games.

AMD still runs cooler, is cheaper, and draws less power as well. Am I missing something?

Basically. They've introduced i7 and i5 chips that are competitive with AMD in the same way the i9 is, but considering you'd have to buy a $100+ AIO to really get performance out of them I'm not sure who those chips are for.

Malcolm XML
Aug 8, 2009

I always knew it would end like this.


HalloKitty posted:

It's OK, but I'm a bit surprised that 10 cores at 4.9GHz burning an enormous amount of power couldn't catch up to a 3900X in productivity tasks. I thought maybe the sheer clock advantage would have helped there.
It really is getting a bit desperate now though, they really need a process shrink or a newer architecture, we're reaching Peak Skylakeô

i guess the 4 extra threads swamps the 10-20% clock advantage

rocket lake is the new arch coming soon (tm)

nobody should be buying these processors they are awfully positioned.

Malcolm XML
Aug 8, 2009

I always knew it would end like this.


zen 3 is coming later this year too so gently caress paying for these hot-rear end chips

Shaocaholica
Oct 29, 2002

Fig. 5E


10th gen still 14nm right?

Riflen
Mar 12, 2009

"Cheating bitch"

Bleak Gremlin

Shaocaholica posted:

10th gen still 14nm right?

10th and 11th.
11th will be Rocket Lake, which is supposed to be a new architecture that was originally meant to release on 10nm. It's being back-ported to 14nm. There'll be no 10nm desktop parts from Intel until some time into 2021.

Inept
Jul 8, 2003

3 olives has a payment plan with his phone company for his cell phone

Riflen posted:

10th and 11th.
11th will be Rocket Lake, which is supposed to be a new architecture that was originally meant to release on 10nm. It's being back-ported to 14nm. There'll be no 10nm desktop parts from Intel until some time into 2021.

TSMC is beginning volume production on 5nm. That's ~172 million transistors/mm, while Intel will still be on 37.5 million/mm. That's crazy.

Shaocaholica
Oct 29, 2002

Fig. 5E


Itís not as crazy as murder hornets and bee balls but Iím not worried.

Cygni
Nov 12, 2005

raring to post



The 10900k is goofy. The 2 extra cores being tacked on with extra latency is very strange. It really is only for those that will only accept THE BEST gaming CPU.

Just like the Ryzen stack, the lower level parts are much more interesting. The 10700kf is fairly well positioned against the 3800X. You pay a $50 premium for a few FPS now, but frequency does tend to help parts age better. I also think the price will sink a bit.

The 10600 is just too expensive. Like everything else both AMD and Intel make, the 3600 makes it all irrelevant. If it was $180? Or even $200 even? Would be a lot better positioned. But who knows how many of these intel really intends to make and sell with their capacity issues.

WhyteRyce
Dec 30, 2001



Enthusiasts shouldn't be buying them but fortunately for Intel enthusiasts aren't the ones with those giant lucrative deals and contracts that keep them afloat

GRINDCORE MEGGIDO
Feb 28, 1985


Wonder how well the 10700kf would overclock.

Paul MaudDib
May 2, 2006

"Tell me of your home world, Usul"


GRINDCORE MEGGIDO posted:

Wonder how well the 10700kf would overclock.

MSI's engineering samples suggest that the 10900K gets basically all of the gold-tier silicon, the 10700K and 10600K didn't overclock incredibly far, they were all in the "about average" or "below average" bins and generally needed more voltage at any given clock.

https://www.overclock3d.net/news/cp...e_series_cpus/1

I don't see anything that changes what I said before, the only chips that look notable in this lineup are the 10900K if you gotta have the best right now, and the 10700F if you can use MCE to get all cores to the max turbo, because that gets you a 4.8 GHz all-core 8C16T for under $300. But in both cases you really should just wait and see what happens with Rocket Lake and Zen3 later this year. Rocket Lake should be a substantial IPC boost and that lets them pull back on the clocks a bit, as well as bringing PCIe 4.0 and 4 additional CPU-direct PCIe lanes, and Zen3 should bring AMD's gaming performance pretty close to where Coffee Lake/Comet Lake currently are.

Paul MaudDib fucked around with this message at 20:06 on May 20, 2020

WhyteRyce
Dec 30, 2001



I remember when I got a free hand-me-down dual core Pentium 4. Those were good times

Budzilla
Oct 14, 2007

I do it because I love you.


Riflen posted:

There'll be no 10nm desktop parts from Intel until some time into 2021.
It's amazing that people keep saying 10nm will be out in 18 months or so after how many years of delay? It's been a disaster for Intel and the only time I will believe on a ready date for 10nm desktop is when I can go out and buy one.

HalloKitty
Sep 30, 2005

Adjust the bass and let the Alpine blast


WhyteRyce posted:

I remember when I got a free hand-me-down dual core Pentium 4. Those were good times

Prescott was hot, and slightly worse per-clock than Northwood. What could be better than slamming two of them together and calling it the Pentium D? Nothing; best to ignore the 130W TDP. Not to be confused with Pentium Dual Core, of course, which was Core based. Still, for naming confusion, that has nothing on 10900X and 10900K. Good job, Intel

HalloKitty fucked around with this message at 20:23 on May 20, 2020

Adbot
ADBOT LOVES YOU

Shaocaholica
Oct 29, 2002

Fig. 5E


I still have a loaded Dell Precision 670 workstation with 2x dual core pentium4 Xeons. That's 4x P4 cores in one system its amazing.

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply
«562 »