|
davebo posted:Well I ended up with the -C since that's what microcenter had in stock this morning, but out of curiosity, would there have been any benefit to the intel ethernet for someone at a house with no other networked computers and only 20mbps comcast cable? It really depends on how bad the non-Intel ethernet is--I've owned motherboards where the onboard couldn't hold a connection after waking from sleep, couldn't handle heavy torrenting, etc. But these days that is much less likely. You'll probably be fine and if you hate it you can get a PCIe gigabit Intel NIC for like $20.
|
# ? Jun 4, 2013 16:24 |
|
|
# ? May 18, 2024 06:39 |
|
TheQat posted:It really depends on how bad the non-Intel ethernet is--I've owned motherboards where the onboard couldn't hold a connection after waking from sleep, couldn't handle heavy torrenting, etc. But these days that is much less likely. You'll probably be fine and if you hate it you can get a PCIe gigabit Intel NIC for like $20. I've used whatever default onboard lan has come on Asus and Gigabyte boards since people started bundling that on the motherboard and the only problem I've run into is on my most recent work pc, sometimes the realtek controller finds an extra unidentified network which somehow causes the internet to not work. When that happens I just disable/enable the adapter and it fixes it.
|
# ? Jun 4, 2013 16:50 |
|
The Atheros(?) built in NIC on my current desktop has four operational modes depending on the driver version and setting being used:
Anyway, back to Haswell. Looking at this table, are the workstation/server processors without IGP going to be the eventual -E series? Considering how many different versions of IGP they're going to have as it is, I'd think that it would also make sense to have IGP-less versions of regular processors, easily saving a third of the total die size, according to this: But an i7-4770K without graphics and at 2/3 of the price isn't going to happen, is it?
|
# ? Jun 4, 2013 16:50 |
|
mobby_6kl posted:The Atheros(?) built in NIC on my current desktop has four operational modes depending on the driver version and setting being used: Yeah, pretty much. If you run into issues davebo, just grab a PCIe NIC, no problem. quote:Anyway, back to Haswell. Looking at this table, are the workstation/server processors without IGP going to be the eventual -E series? Dumping IGP gives you room for more memory channels + cores which is what the E5s & E7s desire. And seeing as more memory channels require more physical I/O, that's the reason you have the giant socket for server platforms. I don't think there's a giant market (maybe Intel disagrees) for dropping a hexacore part into the mainstream desktop socket, vs. dropping it into the "extreme"-tier SKU/platform as they've always done.
|
# ? Jun 4, 2013 16:56 |
|
Anandtech has an article listing the dual-core Haswell SKUs. Based on the specs they are a step back from Ivy Bridge, but in actual testing they deliver higher sustained clockspeeds because they run cooler and don't throttle. This should go hand in hand with significantly improved performance.
|
# ? Jun 4, 2013 20:08 |
|
They also put up an overview of the Asus motherboard SKUs; I'd still stick with Plus or above, personally.
|
# ? Jun 4, 2013 20:28 |
|
Review of five Z87 boards priced under $220 http://www.tomshardware.com/reviews/z87-haswell-motherboard-review,3524.html In a shocker, Asus takes the overall and ASRock the value laurels, though the the performance between the Asus and Gigabyte board were pretty similar the former has more goodies.
|
# ? Jun 4, 2013 20:45 |
|
Interesting that you can apparently hook up Z87 Express to LGA 1155. Also interesting that the DMI 2.0 interface between the chipset and the CPU is starting to look shoddy when put into context: it can only move data to CPU/memory as fast as a fraction of the I/O it offers, about PCIe 2.0 x4.
|
# ? Jun 4, 2013 21:35 |
|
Perhaps Intel mailed it the gently caress in on its desktop CPU line?
|
# ? Jun 4, 2013 23:15 |
|
Yudo posted:Perhaps Intel mailed it the gently caress in on its desktop CPU line? As more and more time passes and I think a bit about just how dickheaded some of the things Intel did this generation are (seriously, the glue is too thick on the IHS and that results in a 20ºC temperature delta! What kind of bullshit is that! And TSX, I wanted to mess around with TSX. And other stuff too. I mean, I get market segmentation. I just don't get the appropriate length for parenthetical asides.)... I have to say it really does feel like they're totally yawning in the general direction of desktop users, trusting the majority of consumers to never know they have a hot running processor needlessly because most consumers won't ever care. Cool poo poo for servers, very cool poo poo for mobile, dick for desktop. And my Sandy Bridge "still kickin!'" is turning into an "Ah, so this is what a monopoly in x86/x64 computing looks like."
|
# ? Jun 4, 2013 23:28 |
|
Right now all my excitement for Haswell rests on how cheap I can get a BGA chip in a mini-ITX board. And the 2nd-hand market for Sandy and Ivy.
|
# ? Jun 4, 2013 23:54 |
|
I'm wondering why there aren't any
|
# ? Jun 5, 2013 00:03 |
|
There will be dual-core versions with GT3 at 15W (Iris 5000) and 28W (Iris 5100). All mobile, now, but maybe desktop ones later. No quad cores (yet), though.
|
# ? Jun 5, 2013 00:13 |
|
A terrible path led me here... Although it did reduce my load temps at 4.5 GHz @ 1.3v by almost 20 C.
|
# ? Jun 5, 2013 03:32 |
|
Nostrum posted:A terrible path led me here... Nice. So 4.5@1.3v is your max? Also, the single edged kind of razor might have been a bit easier on your fingers.
|
# ? Jun 5, 2013 03:34 |
|
Shaocaholica posted:Nice. So 4.5@1.3v is your max? Pretty much. It will boot at 4.6 and seems stable in Windows, but Aida64 crashes it after about 5 minutes. While yes, it would have been, it also gave me extra incentive to be really, really careful.
|
# ? Jun 5, 2013 03:39 |
|
Nostrum posted:Pretty much. It will boot at 4.6 and seems stable in Windows, but Aida64 crashes it after about 5 minutes. Dang. I was hoping on having something more than 4.5 but I haven't bought poo poo yet. What are your load temps like?
|
# ? Jun 5, 2013 03:43 |
|
Shaocaholica posted:Dang. I was hoping on having something more than 4.5 but I haven't bought poo poo yet. What are your load temps like? Maxes out around 68-70 C, with very short peaks to around 80 C (I think Aida64 turns different parts of the core on and off). Average temps were 55 C.
|
# ? Jun 5, 2013 03:46 |
|
Nostrum posted:Maxes out around 68-70 C, with very short peaks to around 80 C (I think Aida64 turns different parts of the core on and off). Average temps were 55 C. I don't know if it uses it but AVX gets Haswell very hot.
|
# ? Jun 5, 2013 04:08 |
|
If I'm not going to be overclocking would you still recommend cracking open the cpu? I mean is the heatsink's fan going to be constantly spinning at a higher speed if I don't? Or are the temperate gains only apparent when you're overclocking it?
|
# ? Jun 5, 2013 05:06 |
|
It's only worthwhile if you're overclocking.
|
# ? Jun 5, 2013 05:26 |
|
davebo posted:If I'm not going to be overclocking would you still recommend cracking open the cpu? I mean is the heatsink's fan going to be constantly spinning at a higher speed if I don't? Or are the temperate gains only apparent when you're overclocking it? No, it's risk for nothing. It may run a little hotter, but heat problems become significant at greater than stock voltage. Further, it's rare that a "real world" application would put the same sort of load that stress testing programs for any extended period of time.
|
# ? Jun 5, 2013 05:27 |
|
If I seem to be voltage limited, this wouldn't help at all would it? If only to keep my temps stable at the voltage required? I need 1.248 for 4.3, my chip is horrid
|
# ? Jun 5, 2013 06:04 |
|
In the Anandtech ASUS video the rep says that the Z87-WS motherboard supports ECC memory [with a E3 Xeon]. Link here. I can't find any reference to this on Asus' website and from what I know about the Z87, this should be impossible. Did he misspeak, or will this board really take ECC with an E3 Xeon? Chuu fucked around with this message at 07:18 on Jun 5, 2013 |
# ? Jun 5, 2013 07:01 |
|
The memory controller is on the CPU. The motherboard only provides traces from the CPU to the DIMM slots. I don't know if that's conclusive, but it makes it seem plausible.
|
# ? Jun 5, 2013 07:19 |
|
Factory Factory posted:The memory controller is on the CPU. The motherboard only provides traces from the CPU to the DIMM slots. I don't know if that's conclusive, but it makes it seem plausible. Just got a response from ASUS: ASUS Support posted:Dear [Chuu], That is awesome news.
|
# ? Jun 5, 2013 09:47 |
|
Chuu posted:Just got a response from ASUS: That really is awesome news. Has this ever been the case before? Ever?
|
# ? Jun 5, 2013 09:48 |
|
I can only speak anecdotally, but my server uses a C202 chipset and an i3-2100, and it's running the DRAM in non-ECC mode even though they're ECC DIMMs. I think this kind of thing has been the case ever since the memory controller moved to the CPU die.
|
# ? Jun 5, 2013 09:56 |
|
If removing the lid creates such a temperature difference, why the hell is Intel putting one on to begin with? They didn't feel it being necessary in the past, either.
|
# ? Jun 5, 2013 11:46 |
|
Combat Pretzel posted:If removing the lid creates such a temperature difference, why the hell is Intel putting one on to begin with? They didn't feel it being necessary in the past, either. It works. It works at stock speeds. They clearly don't give a drat about overclockers. There's also the issue of cracking the die - but so far most gains have been achieved by de-lidding, cleaning all the black silicon that holds the IHS on, replacing the Intel gunk with fresh high quality paste, and re-fitting the IHS. Agreed posted:And my Sandy Bridge "still kickin!'" is turning into an "Ah, so this is what a monopoly in x86/x64 computing looks like." Yup. Arbitrarily missing features from unlocked CPUs, godawful heat transfer AGAIN through the IHS, performance focus by the wayside to focus on mobile. Not exactly a glowing time for enthusiasts. Also, 4.5 @ 1.3v as a de-lidded max is a very disappointing result. HalloKitty fucked around with this message at 12:23 on Jun 5, 2013 |
# ? Jun 5, 2013 12:02 |
|
Combat Pretzel posted:If removing the lid creates such a temperature difference, why the hell is Intel putting one on to begin with? They didn't feel it being necessary in the past, either. Because, as someone mentioned before, you could crack the die while mounting in the heat sink without the lid. The very first PC I built (long time Mac user) was an Athlon XP 1700+ that I promptly cracked while putting on the god awful heat sink my buddy's brother recommended. After letting the magic smoke out, I was off to the store to buy another proc, and the burnt smell reminded me of my mistake for a week.
|
# ? Jun 5, 2013 12:19 |
|
Agreed posted:That really is awesome news. Has this ever been the case before? Ever? The only physical thing the mobo manufacturer has to do is run the extra traces to the slots to support ECC from the socket; a lot may decide not to do this due to routing issues/clearance, etc. After that, it's on the BIOS executing Intel's MRC that sets up the DRAM controller to properly enable ECC.
|
# ? Jun 5, 2013 14:46 |
|
Hold your horses gentlemen, Asus has been really flakey about ECC support on non-Xeon based chipsets for a while. I did research for several days while troubleshooting a troublesome board-memory configuration problem on my S1200BTS motherboard and nearly bought a Z77 board except... 1. In addition to ECC RAM and CPU supporting the ECC circuitry, you need BIOS / EFI support 2. There is evidence that in recent years putting ECC RAM onto Asus boards that aren't C20x class boards (via some probing and diagnostics) that ECC is simply not enabled. 3. Despite (2) I also see in my memtest86+ runs that my own C202 board with certified Samsung UDIMMs also says that ECC is disabled during my tests (but they do run specific ECC patterns anyway it looks like). 4. On a forum I visited, someone contacted Asus support (specifically because Asus has a history of supporting a little more than they advertise) about using ECC RAM, bought it, and found that it wasn't actually enabling ECC on their (I think X79) board. If you want to be an early adopter Haswell ECC cowboy wrangling RMAs and boxes for weeks, by all means go ahead and report back. But please don't accept anything from Asus support as gospel about ECC support on a consumer board of any sort at this point until we have some evidence to the contrary.
|
# ? Jun 5, 2013 15:26 |
|
HalloKitty posted:Arbitrarily missing features from unlocked CPUs How many overclocking stress tests even exercise VMX functionality?
|
# ? Jun 5, 2013 16:35 |
|
Apparently Prime95 has supported AVX for a while. Not sure what the fuss is about Aida supporting AVX which was brought up a few days ago. Or maybe it was some other new instruction. edit: Oh, maybe it was AVX2 Shaocaholica fucked around with this message at 16:46 on Jun 5, 2013 |
# ? Jun 5, 2013 16:44 |
|
JawnV6 posted:How many overclocking stress tests even exercise VMX functionality? I would have been really upset if Vt-x was disabled on K SKUs; I can understand no Vt-d support though, as I'm not running PCI passthrough on my desktop (almost certainly on a dedicated hypervisor box though). No TSX sucks on Ks, but let's face it, business adoption of CPUs that support that + eventual roll out of software to leverage it will all be utilizing non-K SKUs.
|
# ? Jun 5, 2013 17:04 |
|
Shaocaholica posted:Apparently Prime95 has supported AVX for a while. Not sure what the fuss is about Aida supporting AVX which was brought up a few days ago. Or maybe it was some other new instruction. Aida64 does use AVX2, so that's probably what was spiking my temperatures.
|
# ? Jun 5, 2013 18:01 |
|
movax posted:I would have been really upset if Vt-x was disabled on K SKUs; I can understand no Vt-d support though, as I'm not running PCI passthrough on my desktop (almost certainly on a dedicated hypervisor box though). Just thinking of the case where someone OC's, 'proves' it with a compute-heavy workload, then throws a VM on it and tears hair out when a leaf of a rare vmexit becomes the critical path.
|
# ? Jun 5, 2013 19:07 |
|
JawnV6 posted:Just thinking of the case where someone OC's, 'proves' it with a compute-heavy workload, then throws a VM on it and tears hair out when a leaf of a rare vmexit becomes the critical path. And this is Intel's problem? I think overclockers have been pretty good about not blaming the chipmaker even on 'proven' speeds. Its 'proven' until its not. There's no wipe-hands rock solid anything with overclocking. Passed 200 hours of prime95 but crashes on Crysis right away? I've seen that case and never have I heard anyone blame the chipmaker or the developers. Shaocaholica fucked around with this message at 19:17 on Jun 5, 2013 |
# ? Jun 5, 2013 19:15 |
|
|
# ? May 18, 2024 06:39 |
|
Shaocaholica posted:And this is Intel's problem? I think overclockers have been pretty good about not blaming the chipmaker even on 'proven' speeds. Its 'proven' until its not. There's no wipe-hands rock solid anything with overclocking. It's not, it's just a curious edge case where OCers will go crazy because "traditional" stability tests aren't a 100% guarantee (they never were/are). A large team of guys and a pile of money spend a copious amount of time doing verification/validation (search LinkedIn for validation/verification engineers at Intel and see how many hits you get) on silicon and while most bugs are fixed, there's still plenty of errata that makes it out to customers. Either way it's not an issue because those who depend on VMX will never be overclocking and all the turbo boost frequencies are validated.
|
# ? Jun 5, 2013 19:21 |