LOL! that's been my thoughts for MANY moons LOL! I would not be surprised if most cpu's are simply different internal clocks/settings. E.g. the 945, 955, 965 are all the same chip, with different settings. And since I bought the 965, you're all probably laughing at this statement LOL! Well... the things is, if I'm mistaken I certainly want the best
Speaking of 955's, what do you guy's think of this as a budget AMD build? GIGABYTE GA-770TA-UD3 HD 4850 CORSAIR CMPSU-450VX AMD Phenom II X4 955 Mushkin Enhanced Blackline 4GB I'm currently undecided about the ram and it's likely to be changed since the rebate on the mushkin will be well over by the time this build gets done. I think cost could be cut by getting a cheaper gpu but I wasn't sure what to get. This system would pretty much only be used for email, internet, light video playback and unfortunately autoCAD so it can't be removed entirely. TIA, redmaw Edit: I was wondering earlier today what was with the two 955's on newegg, thanks Russ X)
Red_Maw, You're Welcome. Some motherboard manufacturers are going a bit nutso with the layout. Note how close the PCI-E x16 slot is to the Northbridge and the CPU cooler. I guess it it's a two slot video card, the heat inside should be minimal. Shortening the pipelines is one thing, but it seems to me, putting all the hot stuff in one general area, sure makes it more difficult to cool everything well! Russ
Both 955.s are with the C3 stepping and as said on the first reviewer on the Gigabyte board this is one of the quirks with the 955.I have it running at 3.6 on air. Take a look over in extreme and Rebels Haven on the 955 after the problems with the second one I found.My mistake was I should"ve kept the 945 for the +2 board instead of the 955.The 945 overclocked a lot easier than the 955.The 955 liked the Gskill ram better than the corsair.
Thanks for pointing out the layout, I didn't even notice. After a brief look it seems that all Gigabyte boards have the first PCI-E slot up at the top, can't imagine why they decided to do that. Going to have to take a closer look at MB's from now on, and it's making me reconsider Gigabyte.
The top 1x slot, despite how it looks, does not interfere with the northbridge cooler on 90%+ of boards. I found this out recently by using one for the first time.
Sam, It's not the 1x slot that concerns me, it's the top PCI-E slot. There's only about 60mm of space between the back side of the Video card and the CPU cooler, with the card in the top slot, and the NB heat sink sits squarely in the middle of that space. I'm going to put the video card back in the top slot and see what that does to my temps. Best Regards, Russ
I won't lie, upgrading from my HD3870 to 4870X2 in the top slot upped my old NB temps by 14C, and moving the X2 to the lower slot gave me 12 of those Celsius back As long as you have a board with proper electrical connection speed to the lower slot (any high-end board in a series, the basic boards like UD3 ones will be an issue) then using the lower slot is fine, even for a dual card like a 4870X2.
Sam, I think it's the fact that when the card is installed in the top slot, there's almost no air over the northbridge at all, as it's pretty well blocked from most of the airflow over the MB. I'm going to leave the video card where it is. 3Dmark measures the video performance the same, regardless of which slot I install it in, and my GPU temps are 6-7C lower in the bottom slot, as well. Russ
I'm gonna do that as well russ. If for nothing more than to experiment. I really think I'd get better temperatures as well. I'm probably gonna end up replacing the Tuniq 120 fan. I don't like the way it behaves. I think a PWM fan may be in order. Perhaps the scythe that Russ recommended. Though I'd like one with slightly higher RPMs. I'll do some shopping tonight
That depends on how wide the case is, and where the airflow is from the side and the front. If you only have a weak side fan or no side fan, and the only front fan is at the bottom, then this is a big issue (or if you have a very narrow case, and a long graphics card). Having side fans at the top as well as the bottom did wonders for my temperatures with the HAF. Omega: It's worth remembering that motherboard fan controllers control voltage, not rpm. Typically if your CPU is nice and cool it will be about 7-7.5V, at normal good load temperature, around 10-11V, and 50 celsius plus, it'll be the full 12V. If you have a high speed fan, this means it will almost always be noisy. Since 1600rpm (or 1300rpm with 9 blades) is about the optimal noise/cooling ratio for a 120mm fan tower cooler (anything above is lots of extra noise for 1, maybe 2 degrees off) I recommend 1600rpm fans like the Scythe SFF21F, or 1200rpm Slipstreams, depending on the fin spacing.
Oman7, I picked the Scythe I linked you for it's CFM performance and low noise. 1200 rpm and almost 53 cfm will cool anything, but I doubt it would ever run that fast to begin with. I envision 700-800 rpm being about the max it will run. I have a 53 cfm Silverstone on the rear of my computer and it's second only to the 800 rpm Kama bay fan in the low noise department. It should be very quiet. Russ
Off a normal motherboard controller, you will be looking at around 650-700rpm at idle, or if you've got very good cooling and/or a low voltage, perhaps 500. At load, if you get amazing temperatures, perhaps 850-900, normal sort of temperatures 1000-1100, and a bit on the warm side the full 1200.
800rpms sure sounds nicer than what I currently run. Right now, it either hits 200Rmps, or 1600. Rather irritating. 2 speeds! Slow and fast! Nothing in between!
Sam, I have two levels of airflow, front to rear. One at the bottom for the HDDs, and one right in line with the CPU cooler. Having good airflow in that area makes all the difference in the world to having good temps. Oh! The fan I linked to Oman7, is PWM. It should work brilliantly! As long as the case airflow is sufficient, I'm still betting on 700 to 800 RPM max doing the trick with the Tuniq 120. Russ
NO! It's extremely obnoxious. Obnoxiously fast! If I replace its fan, can I expect as good or better results? 1600 is reasonable, ONLY if its under heavy load LOL! But its not. It simply seems to only have 2 speeds. Slow(200Rpms), and Stupid fast (1600-1750).
The fan controller's operation will not change based on a new fan being installed. If the fan goes straight to full speed as soon as you load the CPU, the CPU temperature is too high, and you should either down the voltages, or if you want lower noise, use a quieter fan (which you're doing) and/or use a manual fan controller that's not so strict.
Oman7, I'll make you a deal. Buy the Scythe, and if it doesn't do the job I will buy it from you, full price. I was just going to order one to replace the rear fan and move the Silverstone to the lower front, but I'll wait to see how yours works first. I have a second 4 pin header on the MB that runs PWM, so it should work even better than the Silverstone with these low temps I have now. It's sure nice to have a quiet computer again! LOL!! Russ
I've been in the process of learning what all the adjustments in the bios have to do with overclocking. One of the most interesting things I've learned is about the CPU NorthBridge VID voltage. It controls the communication between the NB and the CPU. The NB voltage only controls communication between the motherboard and the Northbridge. If you raise the NB voltage without raising the NB frequency, all you are doing is creating more heat, and nothing else. Here's my current settings for my Athlon IIx4 630 Quad. The CPU host clock Control is set to manual. The CPU host Frequency is set to 250x14 for 3.5GHz. The CPU Voltage is set to 1.424v (Max). The CPU NB Frequency is using the 9x multiplier for 2250MHz. I've raised the NB Voltage to 1.200v. I've also set the Southbridge Voltage to 1.300v The HT Link Frequency multiplier is set to 9x for 2250MHz. I've set the HT Link Width manually to 16 Bit. And the CPU NB VID voltage is set to 1.350v. The CPU NB VID is the key to getting things stable, and allowing the NB to run stable at 2250MHz, which speeds up the Data flow between the NB and the CPU. It was creating a bottleneck between the faster CPU and the much slower NB (1500MHz slower), and generating CPU errors in the process. The end result seems to be slightly better benchmarks, and a nice fat memory bandwidth of 8000MB/s, and my lowest SuperPi 1M yet at 23.078 seconds. The main benefit is faster throughput for I/O and faster data speed, so internally, it runs noticeably faster! now if I could only get some decent information on the ACC, because AMD has been very closed-mouthed about how to use it, and there is much conflicting advice on the net about it's settings. I do know that when set right, you can up the NB and HT link frequencies as high as 2600MHz. I'm pretty sure that everyone "knows" that the NB frequency and the HT link should be set as close to 2000 as you can get, because that's what we've all been told! All I can say is that's just baloney. Before doing all this adjusting, you couldn't even get the computer to boot if you set the NB and HT link frequencies to 2250MHz on mine, and now it's running that speed, and so far is solid as a rock. The conservative view is 1.400v max, for the CPU NB VID Voltage. Mine seems to be running fine at 1.350v. No heat issues at all, and it's the best it's ever run. As I learn more, I'll post it here! Stay Tuned, Russ