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 money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

How can a mostly fixed function GPU be DirectX 10 compliant?

Adbot
ADBOT LOVES YOU

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

The encoding engine will be nice for video chat and poo poo like that, saving power and such.

But for proper video encoding (like "archiving" movies), it's loving useless. For one there's no two-pass mode, if necessary, like to hit a certain file size, and two, I really doubt that an on-chip implementation can reach the complexity of a software implementation. You know, when it comes to analyzing the image content and deciding how to compress something efficiently in both size and quality.

I really wanted the GPU to have GPGPU capability, to get free physics acceleration. But hey, can't have everything.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

~Coxy posted:

I'm a bit disappointed by the lack of PCI-E lanes and USB 3 support.
Won't the LGA2011 version have plenty of lanes? --edit: A quick google says 40 PCIe 3.0 lanes.

MrBond posted:

I thought Handbrake and x264 shun 2-pass encodes in favor of the "quality based" ones now?
If you need to hit a certain file size, there's no way around two passes.

Combat Pretzel fucked around with this message at 11:03 on Sep 19, 2010

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Aleksei Vasiliev posted:

I don't need either of those. And I'm pretty sure encoding to a constant quality and limiting the max bitrate means you don't need 2-pass for device-compatible encoding, too.
If CBR would churn out decent results, people wouldn't have invented two-pass modes.

And in regards to the Intel Media Engine, I'm pretty sure it won't hold its water to software implementations in terms of encoding quality and complexity.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

On one hand, I'm itching to get a Sandy Bridge. On the other hand, I'm going to sit it out a while, waiting for prices to come down, especially since I expect the LGA2011 variants marked up like poo poo.

I do own a C2Q 9450 right now, so I'm not pressured.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

BangersInMyKnickers posted:

UEFI, I was told anyway, also allows for the operating system to do a warm reboot without UEFI re-executing. So there is that.
That's not really a function of UEFI, altho it might facilitate that.

OpenSolaris could do a warm reboot on its own, as long all drivers could make their devices shut up. The Linux kernel came out with their own possibility of warm rebooting shortly after, altho they still seem to keep it for kernel testing only.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Anything special in the Sandy Bridge LGA2011 over the LGA1156 one? Apart from more PCIe lanes and quad channel memory?

--edit: Also, with apparent support for things like OpenCL, does that mean that games can actually use the graphics unit for physics? Nevermind, no OpenCL.

Combat Pretzel fucked around with this message at 20:12 on Nov 16, 2010

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Raptop posted:

Sandybridge definately supports OpenCL
From what I've read, their OpenCL implementation will use SSE and AVX, not the integrated GPU.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Blargh. If I want the variants with bigger cache, I also get hyperthreading.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Alereon posted:

You don't want to turn HT off, it will improve performance any time you've got more threads running than cores. It may not be worth paying extra for if you don't frequently use heavily-threaded apps, but there's no reason to give it up if you're getting it.
The problem I see is that game performance might get a little negative effect by using Hyperthreading. The Windows 7 thread scheduler is HT-aware, but if for some reason it keeps dispatching secondary threads to an already loaded core, it may affect performance.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

japtor posted:

Is this a theoretical concern or does this happen with the current HT implementation?
Apparently that was the case with the older HT implementations. It doesn't matter during normal usage (web browsing and that sort of poo poo).

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

I hope there'll be Bulldozer benches soon. I'm wary about its shared module crap, but I'm willing to get convinced. Not entirely happy with half the AVX capacity on it, then again, do games even use SSE to its full extent (SSE3 and upwards?)

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

I just noticed that UEFI doesn't seem to be mandatory with the Sandy Bridges. At least the first mainboard I checked on Gigabyte's site mentions a regular Award BIOS.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Gigabyte is to be avoided like the plague. I found various previews of their new boards and they do indeed still have regular old BIOSes.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

spasticColon posted:

Then which motherboard maker is going to have UEFI on their SB boards? And is UEFI needed yet other than allowing the boot drive to be larger than 2TB? I would like to get a board with UEFI but I don't want to pay a premium for it. If they cost significantly more than BIOS boards then I'm sticking with a BIOS board.
All upcoming SB board from Asus do have EFI. And yes, you need EFI for +2TB drives, at least if you don't want to partition it. If the boot partition's below 2TB, then it should be OK I think.

I'm going with the P8P67 Pro from Asus currently. I like Gigabyte, I currently have a loving expensive one from them, but their support and promises are usually bullshit. So if the current series of boards don't have EFI, they'll never get it.

For me it's the principle. Everyone's FINALLY going to use EFI, and they're trailing behind.

Combat Pretzel fucked around with this message at 11:19 on Dec 29, 2010

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Anyone knows whether there are virtualization improvements coming with SB? My Google-Fu leads me always to old Nehalem uArch slides and pages.

--edit: vvv Multiple PSUs with power-on chaining.

Combat Pretzel fucked around with this message at 17:37 on Dec 29, 2010

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Raptop posted:

I think VMCS cache is a feature for the Xeon sku only.
Trying to figure out what VMCS is, I ran over VirtualBox documentation that suggests that it's a feature available on all Intel CPUs with VT-x.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Hmmm, I've been thinking... The does the SB GPU have embedded memory for the main framebuffer, or is it also stored in main memory?

If latter is the case, I hope that various virtualization manufacturers remember that there's an on-die GPU and think about repurposing it. Like this, one guest can actually get hardware accelerated graphics without much issues and messing around. Xen and KVM already have "mostly" working code to passthrough primary and secondary adapters to their guests, I'd guess at least VMware is also toying around with that stuff.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Alereon posted:

The GPU shares L3 cache and system RAM, to my knowledge. Ivy Bridge will change this by integrating up to 1GB of RAM using stacked LPDDR2 dies with a 512-bit bus. Assuming 800Mhz LPDDR2, that's 51.2GB/sec of memory bandwidth for the on-die GPU.
Bleh, so in that case, reusing the built-in GPU for hardware accelerating a virtualized guest is a short lived idea, unless a hypervisor can access the embedded/stacked memory.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

I'm glad that Anandtech mentioned it. I want/need the IOMMU for virtualization purposes. So I'm going to avoid the K-series like the plague. i7-2600 it is.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

The LGA2011 variant this fall is going to have 40 or something lanes.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Not sure how Windows does it right now, but other operating systems use the IOMMU from VT-d for various driver related things outside of virtualisation, for performance reasons.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

If you're deciding on a P8P67 or upwards from Asus and still own an IDE DVD drive/burner, you'll be needing a SATA one. No IDE port on them.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Going for a Asus Sabertooth, Core i7-2600 and 16GB RAM. Everything's here but the CPU

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Kashwashwa posted:

I'm honestly curious to know what people are doing with so much RAM? What could you possible need that much for?
Virtual machines, for one. I'm running a virtual file server on this thing. I'd like to run it with more cache.

And because I can. When I remember what 8MB RAM cost me way back, this is peanuts.

--edit: Yay, found a shop with the CPU in stock!

Combat Pretzel fucked around with this message at 19:04 on Jan 7, 2011

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

UEFI speeds up POST and subsequently boot times, too.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Cryolite posted:

What's so great about the Intel LAN controller found on the Asus P8P67 PRO vs. the RealTek controller on the plain P8P67? Is it just lower CPU utilization?
Yeah, CPU usage.

The RealTek 8111E is a lovely network. It just sends and receives packets the simplest way possible. No acceleration whatsoever, all software.

The Intel controller has send and receive queues, checksum and segmentation offload, interrupt moderation (reduces context switches a lot) and god knows what other stuff to accelerate various TCP/IP tasks.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

So, again, Sandy Bridge and 1.65V memory. Is this OK, if I'm not going to overclock?
--edit: I might try to undervolt it 0.05V to 0.1V, if necessary.
--edit2: These are Corsair DDR3-1600 sticks. I've now read suggestions that they might run just fine at 1.5V if clocked at 1333.

Combat Pretzel fucked around with this message at 14:27 on Jan 12, 2011

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Around here, P67 mainboards still aren't available anywhere. Only these lovely H67 ones.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Siroc posted:

When I boot and check out the BIOS, the ASUS UEFI screen usually says the CPU is at 45C. When I boot into Windows and use their utility, idling is at about 28C. What's the discrepancy?
When Windows is idling, it makes the CPU run HLT commands, which shuts it almost down completely. The UEFI BIOS probably doesn't have a sophisticated task/thread management and keeps running a loop polling for keyboard and mouse events, effectively keeping the CPU busy as hell over nothing.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Well, I finally got a P8P67 EVO from Asus, plugged my DDR3-1600 1.65V 9-9-9-24 memory sticks in and when starting up, it configured it to 1.5V DDR3-1066 and 7-7-7-20. Strange stuff. I manually put it to DDR3-1333 1.55V 9-9-9-24, let's see how stable it is.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

I have the stock cooler on the i7-2600. There's that tool called Core Temp. When running the OCCT CPU test, the temps shoot up to 90C. Are these readouts normal, for CPU internal sensing?

Also, the Asus setup tool has hilariously bad start up times.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Sir Nigel posted:

Provided the readout is accurate that is way way way too hot, especially if its stock clocked. Same questions apply to you as R1CH.
Talking core internal temperatures (apparently there are sensors in the CPU die itself), not the mainboard sensor. When idle, core temps are like 35C.

Stock cooler without anything additional.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Hrm, I think I'll go check the heatsink then.

--edit: I guess the TIM had to melt and distribute itself properly first... Now it stays in the 50C under full load.

Combat Pretzel fucked around with this message at 20:04 on Jan 14, 2011

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Asus already has BIOS updates out for their P67 boards. You'll have to download them by hand tho, since that lovely AiSuite doesn't find them on its own.

FYI, the update resets all settings to default.

--edit: And do yourself a favor, never install the Asus tools. It shits your system up with tons of autostart apps.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Hmmm, I was greeted earlier with the clock interrupt BSOD when I came back home today. Seems like this issue is mostly voltage related when overclocking from what I've read. I think Asus' VRM, EPU and whatever else three letter acronyms are playing against me.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

What I also don't like how the BIOS makes its settings changes. It shuts down the computer, starts it for a few seconds, shuts it down again and then does starts again and does the POST.

That's really nice for hard drives.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

incoherent posted:

Turn off Virtualization in the bios.
It's related to that?

I do actually use virtualization, so I have to live with it. I guess.

Also, MSI has games in their UEFI BIOS

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

I'm running 1.65v RAM at 1.54v at the moment. Intel says the safe range is +/- 5%, which would be 1.575v at the upper bound.

Altho reading out the SPD, it says 1.5v everywhere except for the XMP profile for DDR3-1600, so I could probably try running it at 1.5v safely.

Adbot
ADBOT LOVES YOU

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Yeah well, I had some sketchy plans of going LGA2011 anyway...

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply