|
redeyes posted:also, Marvel Yukon PCI-E Gig nics loving rock. You can get a nice Rosewill one from newegg for 25ish bucks. Didn't Marvel buy some (all?) of that business from Intel? Or was that XScale? Gunder posted:If you're not planning on doing any overclocking, is there any point at all in getting the K versions of these chips? I was planning on getting a 2500. I have been told: "Overclocking the thing will be dead simple if you're not trying to max it out, and the cost increase is so minimal I personally think it would be worth it. It's only $11 more, after all. Any particular reason you're opposed to overclocking?". The thing is, I've always thought that overclocking gives you minimal performance at the cost of stability. Is this no longer the case? Am I a fool for not getting the 2500k? Of course, if your application demands VT-d and TXT, then the choice is forced upon you. graph posted:No, I don't think it's that. Also, lots of Intel goons ITT.
|
#
¿
Jan 3, 2011 21:42
|
|
|
|
| # ¿ Apr 16, 2024 13:07 |
|
|
LooKMaN posted:Do you have any links to reviews or proof for this? I have some cheap Attansic onboard gigabit card on my old C2D motherboard but i've never had any issue with torrenting. But i'm currently only on some 20Mbit connection and i'll be switching to cheaper 100Mbit which might change it, so Iim interested if there's any gains for paying 30$ for an Intel NIC. I'll try to dig up what I can, but to be perfectly honest, my bias towards Intel cards stems from their awesome performance in a few ipcop and pfSense routers I've built (and reports of it on mailing lists), and the lack of awesome performance from Realtek, Broadcom and Atheros add-in or integrated cards in the same systems. I know that there was a length thread on the Atheros sucking on my current mobo (P5Q Pro if I remember correctly), so I purchased an Intel PCIe GigE NIC, and gained jumbo frame support, and stopped losing connectivity intermittently whilst torrenting. I also stream everything from a NAS, so maybe I was just abusing my card more than the average user. I still think folks should go for it if the price difference is negligible/you get other goodies for splurging (like more SATA, SLI, etc). @punk rebel ecks: newegg has always price gouged, I don't remember how much E6600s were when they came out on the egg, but I do remember waiting like 3 months to get one. I think the Ks will see more of a premium, because newegg knows enthusiasts will pay for it and that a lot of people will want one
|
|
#
¿
Jan 3, 2011 22:15
|
|
|
Lief posted:Could somebody break down for me what VT-d and TXT are and how they affect system performance? I've tried looking it up for myself but the amount of jargon is enough to make a legislator blush. If you have to ask what TXT is, it probably isn't for you. From what I've seen, it's really geared for business use, and really shines when some disgruntled employees steals a desktop (or loses a laptop) and the data on the machine hopefully remains safe. Pretty sure it needs a TPM to even function. VT-d is a virtualization extension that boosts performance of guest VMs by giving them among other things, DMA access to host hardware (basically the ability to remap DMA and interrupts). I want to say that this is most helpful when you're running a bare metal hypervisor like ESX, but I'm not 100% sure about that. I don't think the average enthusiast would miss either of these (especially not TXT) in return for having an unlocked multiplier; I know I won't. ^^ VT-x is the generic virtualization tech that Intel intro'd to match up with AMD's AMD-V. It gives you extended page tables from the CPU, stopping the VM monitor from having to handle page faults. VT-c is for network stuff, I don't know too much more about that. An Intel brief I saw split them up thusly: CPU is VT-x, Chipset is VT-d and Network is VT-c. It was some Xeon 5500 virtualization brief. e: You don't need any specific hardware extensions to virtualize. Having them could greatly enhance your VM performance, however. movax fucked around with this message at 00:54 on Jan 4, 2011 |
|
#
¿
Jan 4, 2011 00:48
|
|
|
Octopode posted:Well, for what it's worth, I was just at Fry's here in San Diego, and they had the Asus P8P67 Evo and Pro in stock. I picked up the Evo for just over $200, but the Pro was priced reasonably below that, around $140 if I remember correctly, so you should probably be able to pick up the base model P8P67 for $100 or just above. I wish I had Fry's! I plan on getting the Pro, I wonder if Fry's.com or outpost.com has them... The base P8P67 isn't bad, Legit Reviews got one. They said it cost $160.00 though.
|
|
#
¿
Jan 4, 2011 03:56
|
|
|
Wedesdo posted:The P8P67 has a 16x/4x split for the PCIe lanes. And if you use the 4x electrical slot, all the 1x slots get deactivated. Hm, that's right, it doesn't have the PLX PCIe switch to offer up some more lanes. PCIe lanes are in such short supply  Still, for a single GPU user that wants a decent board, should be fine (along with its Gigabyte rival). I can imagine someone having a single GPU and a x4 RAID card or something, or maybe single GPU + x1 NIC, x1 sound card.e: Oh good, there are still PCI slots!
|
|
#
¿
Jan 4, 2011 05:03
|
|
|
MTW posted:Oh, ok. Am I losing performance by giving up VT-d? I often have two or more virtualized machines running. Theoretically, yes, you are. What CPU do you have at the moment, and what virtualization software are you using? For example, if you have a i7-860, that supports VT-d, so by going to a 2x00K, you may see decreased performance if your VMs make heavy use of host hardware like ethernet or a RAID controller. For PCIe lanes, if we look at this very popular piece of Intel marketing, you have 16 lanes coming from the CPU, and 8 lanes (root ports) coming from the PCH. My handy Ibex Peak datasheet says nothing about reusing the link from the GigE MAC as an additional PCIe link, so I'll just assume that doesn't change and if you don't attach a PHY there, you are SOL. (The link is electrically PCIe IIRC, but it will only function for chipset to talk to a -79 PHY, if I remember my part numbers properly). So the first 16 (PEG0) is sacrificed to graphics. My company happens to not give a poo poo about graphics performance (industrial PCs), so we're very happy to have 16 lanes to hook up to FPGAs and stuff. For regular users though, you either feed this to one GPU @ x16, or 2 at x8 a piece. The remaining 8 are up for grabs. Not using the integrated MAC on the chipset? Poof, one gone to Realtek. USB 3.0 support? Poof, another one bites the dust. Extra 6Gbps SATA from JMicron? Another port gone (JMicron's site is poo poo, I can't confirm if it's x1 or x2) Each group of 4 can become 4 x1, 2 x2, 1 x2 + 2 x1 or 1 x4. So looking at your average mobo that could support at least two GPUs plus other goodies: x8 - GPU1 x8 - GPU2 x1 - NEC USB 3.0 x1 - JMicron JMB362 (2x eSATA) x1 - Marvel 9120 SATA 6Gbps Controller x1 - VIA VT6415 PCI-Express to IDE Bridge x1 - Realtek Ethernet x1 - PCIe to PCI bridge = 23 lanes consumed x1 - Expansion! = additional lanes are required (24/24 consumed) Luckily, Firewire can be on the PCI bus (the VIA chip Asus chose to use is PCI). I assume that all the smart overclocking/embedded controller/power management lives on the LPC bus. So the higher end boards feature sometime of PCIe switch (PLX, Lucid, whoever) and have the nice little caveats of certain PCIe ports not functioning when others are in use, or limited electrical link width on x16 slots. I know Asus has x16 slots that are *'d with 'limited x4 electrical link width'; it's probably PCIe root ports 1-4 or 5-8 ganged together. Likely through BIOS setup you can adjust chipset straps and thusly reallocate PCIe lanes if you need 4 x1 rather than 1 x4. If you are interested in reading about PCIe switches, PLX is a popular manufacturer. For example, take a look at the PEX8608, a 8-lane PCIe switch. Someone could use this switch to feed a bunch of x1 devices (PCI bridges, USB 3.0, SATA, etc) and retain a wider x4 link for another GPU. The product brief isn't under NDA and will provide a pretty decent overview of what PCIe switches are all about. And I guess another fun tidbit about PCIe: while it benefits from being a high-speed serial link that doesn't suffer from clock skew like parallel buses (PCI) did, there are still quite a few traces to layout. At a minimum, a x1 link would have Rx, Tx plus a Reference Clock. Each of these are a differential pair, so you have a Rxp and a Rxn. So that's 6 traces that have to be impedance controlled/length matched for a measely x1 link (txp0, txn0, rxp0, rxn0, refclkp, refclkn). A x16 link would be 16 lanes, 16 transmits, 16 receives + 1 reference clock. That's 66 traces total (ignoring grounds, power, etc). Increasing the number of layers + using blind/buried vias makes these easier to route...but also drives up board cost. Then again, Taiwan is pumping out thousands of these, but that cost still gets passed onto us! e: the reference clock is provided for devices to perform clock recovery on the serial data. They don't care about the phase of the clock, just the frequency. When you turn on Spread Spectrum clocking, it varies this frequency (nominally 100MHz) somewhat (I think as "low" as 99.5MHz), which comes in handy when you're trying to meet EMC requirements from the FCC. movax fucked around with this message at 17:28 on Jan 4, 2011 |
|
#
¿
Jan 4, 2011 17:17
|
|
|
MTW posted:Q6600 and Vmware Workstation. I do make heavy use of the ethernet controller in my vm(s). It looks like the Q6600 doesn't have VT-d. I don't feel qualified to advise you on a purchasing decision solely based on this feature (you might have better luck in the Virtualization megathread), but part of feels that the higher clocks + hyper-threading of SNB would make up for some overhead lost by lacking VT-d. Definitely try and e-mail VMWare before sinking money into what's now "last" generation tech. (In this case, last generation still being a fuckton of CPU horsepower. Even Conroes can still handle most everything thrown at them.)
|
|
#
¿
Jan 4, 2011 21:11
|
|
|
SynVisions posted:I see a lot of people harping about this, but I haven't seen any sort of explanation/estimates of the performance implications. Well, you get a die shrink, along with all the benefits that will bring. The memory bandwidth I feel you'd need to be a real heavy user that thrashes your memory (or a server/computation application) to truly appreciate. Isn't the performance different between running in dual-channel vs single-channel mode in the single-digit percentages in synthetic benches? Z68 will let you use your integrated GPU + overclock like a madman however, so you can look forward to that if you wait. Still a little light on PCIe lanes, however.
|
|
#
¿
Jan 4, 2011 21:16
|
|
|
Dr. Gaius Baltar posted:I wish that review sites would hurry up and do a P67 motherboard roundup. Sandy Bridge goes on sale tomorrow, and I need to know which motherboard I should be buying for my 2500K, and which ones are horrible because they lack UEFI, and/or are hobbled by an insufficient number of PCIe lanes so if you plug something into the x4 socket, the x1 sockets stop working. Plus regular motherboard stuff like stability issues and overclocking ability vs the competition. You could aggregate the reviews yourself; HardOCP looked at MSI, Anandtech looked at ASRock, Legit Reviews looked at Asus (the P8P67 and P8P67 Deluxe, the two polar ends of their lineup, ignoring ROG). I'm sure the usual suspects have looked at Gigabyte and others as well. $178.18 shipped isn't terrible for a mobo, IMHO. It's what I'd expect to pay for a upper mid-range board. I'm planning on buying that board myself. If you don't need SLI, you could drop down to the P8P67. Just don't get the LE.
|
|
#
¿
Jan 4, 2011 22:15
|
|
|
Dr. Gaius Baltar posted:Doesn't the P8P67 have the problem of a x4 slot which, if used, disables the x1 sockets? Update: Looks like the x4 slot defaults to x1 in the BIOS, and if you set it to x4, then it disables the x1 slots and half the USB3 ports. Yes, though it's less of a problem and more of the best compromise they could make given the number of lanes available, which I broke down a few posts above (and guessed at the x4 vs x1 config). To avoid that problem, you have a PCIe switch, which Asus provides to you...for extra money, which is logical. Really you just have to hope for Intel making additional PCIe RCs/lanes available on future chipsets, or filtering down insane numbers of lanes like the X58 had to mainstream. If you look at the list of lanes used I posted: movax posted:x8 - GPU1 and you strip out the extra eSATA, SATA 6Gbps, IDE and Ethernet, you just got back a full 4 lanes to use for other stuff. USB 3.0 is in demand however, so that will present, and Realtek controllers are still apparently cheaper than the Intel PHY, so those two will probably always be present. The reason they threw 3 PCI slots on there is because they couldn't use the space for anything else. We could have the mobo makers say "gently caress legacy, go expansion" and give us something like: x8 - GPU1 x8 - GPU2 x1 - NEC USB 3.0 = 17 lanes consumed out of 24 various combinations of expansion slots that add up to 7 lanes = additional lanes are required (24/24 consumed) If you dropped out the USB 3.0 controller, you could have x8, x8, x4, x4 and nothing else for your crazy SLI/CrossFire config. I'd be happy to have expansion slots instead of having extra 6Gbps SATA, eSATA, and IDE forced on me. Going with an Intel ethernet solution would give back an extra lane that I might not mind giving up to a PCIe<->PCI bridge, because I've still got some PCI cards I use. Yeah, it's not the best. The 2nd x1 slot is really useless because most decent GPUs eat up two slots due to cooling. The 1st you could throw a NIC into though, if needed. It looks like a long card would fit into that slot, as long as you don't mind removing it to get to your RAM sticks. @Alereon: Thanks for the heads up, 4x4GB maybe, here I come! movax fucked around with this message at 00:05 on Jan 5, 2011 |
|
#
¿
Jan 5, 2011 00:01
|
|
|
So, not quite the epic pricing Alereon told us about (probably because these are 4GB DIMMs), but how is the price on these? Thinking about ordering two of them.
|
|
#
¿
Jan 5, 2011 01:38
|
|
|
Alereon posted:I'd buy DDR3-1600 for Sandy Bridge, since it's supported now without overclocking, but that's me. For DDR3-1333, you can get 8GB (2x4GB) G.Skill Value Series DDR3-1333 for $64.99 with free shipping and no rebates. This is the daily Shell Shocker deal, but is a little preview of prices to come. Good point; I was thinking since we are mult based OC now, no need to worry about memory headroom. @Fuzzy Mammal, I think 4 slots is fine because you're not going to be putzing around with CAS/RAS/etc timings + memory speed anymore.
|
|
#
¿
Jan 5, 2011 02:35
|
|
|
The Raglay posted:How are ASRock boards, by the way? I read Anandtech's review, and, especially for the price, I was really digging it. I have a cooler that uses the 775 socket, so I'm a fan of the legacy support, but I don't want to get stuck with a bum board just to save a few bucks. Never used 'em, but they always struck me as a "budget" brand. I'd take Anand's word as gospel though, so if they recommend it, I would go for it. I'm pretty superstitious when it comes to mobos (just like hard drives), so I stick to the brand that's treated me the best, Asus. Alereon posted:There's still the issue that using 4 DIMMs may mean you need to back off on timings or clock speeds for stability due to the additional load and reduced signal quality, and that you can't upgrade your RAM in the future without replacing sticks. I'd say 2x4GB is the best option right now unless you have a compelling reason to get 4 DIMMs. True, but your post about DDR3 pricing made me uber eager to buy it all know instead of getting another 2x4GB sticks 6 months down the line for even more, heh. I suppose if it fucks with my overclock at all, I'll just drop to 8GB and sell the other sticks at a loss. Buying the higher tolerance/speed memory should help out though.
|
|
#
¿
Jan 5, 2011 02:56
|
|
|
greasyhands posted:Asrock is Asus' budget line. quote:They're not really that low end, their motherboards have been getting praises for the performance/features they bring for a cheaper price compared to Asus/MSI/Gigabyte. They're cheaper but not bad quality as you'd expect. R1CH posted:Damnit, that newegg RAM sold out right as I added to cart there will be a new deal next week won't there?
movax fucked around with this message at 05:17 on Jan 5, 2011 |
|
#
¿
Jan 5, 2011 05:14
|
|
|
punk rebel ecks posted:So in short: I'd say Intel would be at the very tip-top for stability. I think eVGA might still make mobos as well. MSI seems to be decent enough to play in the same league as Asus and Gigabyte. Alas, poor ABiT...rest in peace 
|
|
#
¿
Jan 5, 2011 05:40
|
|
|
punk rebel ecks posted:Jetway<<<<<<<<<<AsRock<<Asus =Gigabyte=MSI<<Intel Sure (though Alereon's post feels like validation for my anti-Gigabyte bias, hah). Anand seemed to like the ASRock motherboard they reviewed, just a few minor nitpicks (like no dual ethernet...but when is the last time you used that?). Seems like it will be a real steal. It's definitely better than Asus's budget crippled LE version of their P8P67 family. Also you can toss Biostar/ECS/PCChips on the lower end as well. e: my DS3 LGA775 gigabyte board was also DOA
|
|
#
¿
Jan 5, 2011 05:53
|
|
|
Hardick Hertzer posted:Tech Report put up a nice review & test of four major brand SB mobos yesterday, it's a good read. Nice! Sad to see no UEFI on the Gigabyte board, but looks like there are still a huge number of options present. The Intel board is surprisingly shiny though (and a sane slot layout :yay:)
|
|
#
¿
Jan 6, 2011 18:59
|
|
|
Factory Factory posted:Speaking of RAM deals, I just went in on this one: 8 GB (2x4GB) of G.Skill DDR3 1333 for $74.99 after promo code. According to some Googleable guy on another forum, G.Skill confirmed this stuff (Sniper brand) uses different ICs than the Ripjaws brand, but it's a new product line. G.Skill tends to review well on Newegg, though those brand names are a bit I was totally going to jump on those, but they sold out before I decided too, hah. My only complaint with G. Skill the last time I had their products was the RMA process being a pain in the rear end. But they've recently got a US RMA Center, so it's a lot better now. I'm not an Asus shill (I swear), but the P8P67 Pro is pretty, well...pro.
|
|
#
¿
Jan 7, 2011 06:34
|
|
|
dud root posted:Someone on another forum mentioned that the Asus P8P67 EVO had better voltage/power regulation than the PRO, but they both have 12+2 VRMs. Is there any truth to this claim? No, their VRMs should be identical. The LE and Deluxe are the only ones with unique VRMs (the LE has 6, the Deluxe has 14+2 or 16+2, one of the two). However, Intel themselves only tossed 6 on their motherboard. The extra ones will come in handy when you raise VCore and suck down more power on really high overclocks. @Kashwashwa: alereon mentioned last page that DDR3 prices are about to hit the lowest they will (likely) ever hit; some people will just buy 16GB now when it's cheapest instead of waiting to see when they needed. You could probably kill off your pagefile with 16GB too. @oversteer: definitely do P67 + some AMD GPU. Best bang for buck GPU performance, and coupled with an i5 should smoke most everything for the rest of the year without you having to murder your settings. @Laserface: you might want to look at the Intel, Asus or other boards that sport a nice EFI bios. I think that round-up said the GB board was the priciest of them all?
|
|
#
¿
Jan 7, 2011 16:26
|
|
|
ilkhan posted:If you are that conflicted, and willing to wait a month, wait for the Z67 (Z68?) boards to come out. Best of both in one package. Z68, and I thought it was Q2 at earliest, so it'd be like April or later.
|
|
#
¿
Jan 7, 2011 18:13
|
|
|
Zhentar posted:And it was wrong a decade ago, and has become increasingly wrong since then. Not having a page file will make things slower, even if you have ridiculous amounts of RAM and nothing actually needs to be paged out. Ooops, my bad. I still keep my pagefile around, I just moved it to not-C.
|
|
#
¿
Jan 7, 2011 23:20
|
|
|
zachol posted:How much information is there about Mini ITX motherboard offerings? Asus apparently has mini-ITX variants of their P8P67 family coming "soon". I'm sure Intel will put out a mini-ITX board as well.
|
|
#
¿
Jan 8, 2011 07:04
|
|
|
Alereon posted:TechReport found almost no performance difference between the Intel and Realtek network controllers. Throughput was the same, CPU usage averaged ~3% on the Realtek controller, ~2% on the Intel controller. And this is transferring at full Gigabit speeds. So the differences are basically drivers, and not losing a PCIe lane if you choose Intel (so that lane can go to something else). Ordering 2600K+ P8P67 Pro today. e: why is the LE out of stock?! Poor bastards who bought it e2: I don't even remember how to take the motherboard tray out of my Lian-Li  e3: And I have to find a mounting kit for my Ultra 120 to 1155. 1155 and 1156 have the same holes, correct? movax fucked around with this message at 18:36 on Jan 9, 2011 |
|
#
¿
Jan 9, 2011 18:31
|
|
|
spasticColon posted:Is your case a PC-K7B? If so, what's up PC-K7B buddy? Hah, I wish, I got the  of Lian-Lis, the B70. I just really like case doors  I think the ASRock does pretty well overclocking as well, check out that round-up someone linked a bit earlier. I got the Pro myself, $160 is a price I don't mind paying for a motherboard, considering I keep them for ~4 years before upgrading, and I never throw parts out or sell them (they just get passed on to the next system in line which is why I have retard over-powered file servers and such).Ordered all the parts ($700, I wasn't eating this month anyways...) and found a 1156<->775 adapter for my Ultra 120. Middle of this week, yeah! 
|
|
#
¿
Jan 9, 2011 21:59
|
|
|
SpaceAceJase posted:So what's a good motherboard choice if you ARE going to do SLI/Crossfire? You're always guaranteed a 2 x8 split off the CPU. Without using other chips, the biggest other links you have are 2 x4s from the PCH. That'll almost (never) happen because giving up 2 x4 lanes from the PCH leaves 0 for any peripheral devices.
|
|
#
¿
Jan 10, 2011 04:21
|
|
|
frunksock posted:My understanding was that the manufacturer specifies the voltage necessary to hit the timings they advertise. So I stay away from higher-voltage RAM because it's probably a shittier part that they're compensating for by increasing the voltage, all other things being equal. Sometimes you had to kick up voltage on RAM for stability purposes, especially on performance RAM like Ballistix Tracers or Dominators. With SB however, you are correct, there is no FSB clocking anymore. Just multipliers, and bumping up BCLK (100MHz), but bumping BCLK can very quickly introduce instability into the system since all clocks come from that reference.
|
|
#
¿
Jan 10, 2011 06:35
|
|
|
Paino posted:Oh. No amount of overclocking is possible on the H67? Wow then it's basic as gently caress and while cheap it's probably overpriced anyway. No amount of CPU overclocking, anyways. It's pretty much perfect for OEM builders of consumer and business desktops though.
|
|
#
¿
Jan 10, 2011 15:50
|
|
|
Kashwashwa posted:This is the exact combination of CPU, mobo and Heatsink I just ordered, with approximately the results I was hoping for. I will go back to water-cooling if my particular CPU can sit around 5GHz stable @ <= 1.425V or so. That's just too much e-peen to give up.
|
|
#
¿
Jan 10, 2011 16:12
|
|
|
R1CH posted:Will I have any issues with Windows 7 on this new hardware? I'd really like to avoid reinstalling, will going from a BIOS to EFI have any big impact on this? You should be a-ok, Windows 7 (and Vista) are hilariously tolerant of hardware changes. Might just have to reactivate. My Vista installed lived through 3 different chipsets before I upgraded to Win7. (650i, 680i, P45) @Paino - terrible for SLI since they didn't split the lanes from the CPU. I dislike Gigabyte, but that board seems to be a solid choice for single-GPU, no deluxe-features needed type computing.
|
|
#
¿
Jan 10, 2011 17:28
|
|
|
Paino posted:Thought so. My idea is to put a gtx 570 in it and then, in a poor attempt at future proofing, buying a second one for the SLI when it gets cheaper. However this implies overclocking the cpu a bit, buying an additional fan, upgrading to a better, at least 850w PSU. What resolution do you game at? If you plan to do SLI in the future, definitely get a board that will amenable to it (i.e. 2 x8 Gen 2 lanes split off the CPU).
|
|
#
¿
Jan 10, 2011 20:04
|
|
|
Comatoast posted:These processors/chipsets are supposed to use less power than previous generations. Does overclocking a 2500k negate any of those power saving features? Yes; you are increasing multi and therefore increasing clock frequency, and if you bump VCore for stability, you are sucking down more power (P = IV). The power-saving features will still function (probably), i.e. SpeedStep and such, but yeah, you're drawing more power.
|
|
#
¿
Jan 10, 2011 21:28
|
|
|
Comatoast posted:Thats the one I was thinking of. If the processor is idle it clocks it down to some fraction of its normal clock speed. That one is important. I don't mind a little extra voltage needed constantly but if it never clocks itself down, then overclocking isn't worth it for me. I don't know how it works on Sandy Bridge, but with C2Ds and such it clocks down to some pre-set multiplier values. Since the FSB would be overclocked, your lowest power state would be running faster, because now you would end up with some multiplier times a higher frontside bus clock. I would think on SNB since mults are entirely unlocked, your lowest power state would remain the same regardless.
|
|
#
¿
Jan 10, 2011 23:26
|
|
|
Avalanche posted:Any idea when we should start seeing Sandy Bridge cpus in notebooks? Or will the laptop crowd have to wait until the die shrink with Ivy Bridge? I swear Lenovo announced their lineup with Sandy Bridge CPUs at CES. It's a relatively safe assumption that Apple will have Sandy Bridge Macs this summer also (13" MBP w/ SNB wooo). It's honestly a bigger deal for the mobile market than it is the desktop.
|
|
#
¿
Jan 11, 2011 07:03
|
|
|
Triethyl posted:That 2x4GB pack of G.Skill DDR3 1333 on Newegg is down to 79.99 from 89.99 with promo code EMCKJJJ84, for anyone who didn't already buy 16gb... Fffffffffffffffff... Crossing fingers here hoping none of my parts are DOA.
|
|
#
¿
Jan 11, 2011 15:43
|
|
|
Dr. Gaius Baltar posted:Are there any hard numbers out there indicating that there's no performance loss when using the slower DDR3-1333 memory? Running the slower memory at its (slower) rated speed does lower your memory performance yes, but you'd need a synthetic benchmarking tool to notice any difference, IMHO. If you don't care about winning SiSoft Sandra/etc benchmarks, don't worry about it.
|
|
#
¿
Jan 11, 2011 18:06
|
|
|
LooKMaN posted:Oh god i ordered 2x of that exact same kit and for P8P67 Pro too. It better work or i'll be pissed. I got twin F3-10666CL9D-8GBRL, I hope at least 2 of the sticks are error free so I can enjoy system whilst waiting for RMA. At least with memtest, you can knock out some stability testing whilst testing your RAM too. e: Got $20 credit back (ordered the RAM before promo code) 
movax fucked around with this message at 20:36 on Jan 11, 2011 |
|
#
¿
Jan 11, 2011 20:33
|
|
|
Dr. Gaius Baltar posted:As an update to the P8P67 Pro and memory, it seems that there's nothing wrong with using your DDR3-1600 memory at DDR3-1600 speeds, as long as the memory isn't defective like mine is. I narrowed down the faulty memory module and removed it, and now my computer is chugging along at DDR3-1605. Why 1605 and not 1600? No idea. I set it to DDR3-1600 in the BIOS, and it spat out DDR3-1605, CL9-9-9-24 1.5v. Probably not a big deal. Might be from a slightly overclocked BCLK (100.3, 100.4MHz?)
|
|
#
¿
Jan 12, 2011 03:14
|
|
|
Parts arrived today, can't wait to go home! FrozenCPU shipped my heatsink adapter to the wrong address, now I have to wait until Friday 
|
|
#
¿
Jan 12, 2011 22:27
|
|
|
Factory Factory posted:Yeah, that's pretty high. At 4.4 GHz, I idle at 35 and only get to 70 during stability burn-in, ~60 when doing Folding@Home. I wouldn't go above 1.4V myself. Maybe if I had water cooling, but even then I'd be leery.
|
|
#
¿
Jan 13, 2011 04:28
|
|
|
|
| # ¿ Apr 16, 2024 13:07 |
|
|
kimcicle posted:Can somebody confirm heat sinks that fit LGA 1156 will also fit on 1155? yes. Usual caveats about running into obnoxiously huge motherboard/RAM heatsinks applies.
|
|
#
¿
Jan 13, 2011 15:46
|
|