X58 uncore power draw?
X58 uncore power draw?
The Rampage is designed for quick performance and comes with some interesting upgrades. I used an LN2 jumper to fix the sag. The Giga model looked more impressive than functional, while the Rampage focused on aesthetics. I’ve always enjoyed reliable high-end Asus products.
Some options stand out because they offer a reliable choice without breaking the bank. Most folks settle for X5675s since they’re a solid, decent option compared to pricier models like 5690s.
I also owned a p6t deluxe v2 before switching it for cash and an older MSI H55 that failed after I turned it off in BIOS. It performs better than my average p6x58de, which can't reach 210bclk on any chip since it hits around 220bclk on Westmere. I've even pushed it to 250bclk with slowmode, while the p6x58de barely exceeds 230 in slowmode. I still prefer avoiding ASUS unless I'm using X/E Xeon chips, because the uncore ratio limits RAM overclocking. Gigabyte is also fast but tends to disappear quickly since it lacks a forced uncore ratio, especially on Nehalem boards. I’m also interested in Giga Ex58 boards—they only have four slots, which I can get for $33 when paired with an X5680. There’s also a budget UD4P for $22, and a G1 Assassin for $54 with an i7 970, though I’m unsure about its RAM support given the chip spacing.
It's clear why I steer clear of those x/e xeons. Their performance is limited to around 10x multipliers, so I need a high clock speed to get decent RAM speeds. That's why I'm considering Xeon W or G7 i7 models. For testing, I'd prefer to try several chips myself if possible, rather than relying on sellers. I'm more interested in maximizing RAM frequency instead of pushing CPU speeds, since that's something that often gets overlooked but can be quite impressive. A good cooler or some high-end components like LN2 could really help. If I'm aiming for top CPU speed, I might skip those and go for Westmere quadcores instead.
The focus is always on running high BCLK. Most of the heavy work on X58 happens through big BCLK multipliers, which sets it apart from other systems that mainly rely on CPU/RAM multipliers unless you're using top-tier XOC components. When I set up my X58, I use BCLK and core multi to push the CPU to its maximum, keeping BCLK stable while only adjusting the memory multiplier to lower it back near base frequencies. RAM is only boosted once the CPU is steady. BCLK also naturally rises and falls on most boards; instability occurs but it recovers when you increase it further. This approach was what drew me in, because it offers a fine balance instead of just changing voltage and slamming a single multiplier. From this discussion, it looks like the uncore-mem ratio is a hardware constraint, not determined by the motherboard or CPU.
the system chooses whether to enforce the ratio or not. I can test 2400 ddr3 with 3800 uncore on my x58a udr3, but it crashes in prime95. It still boots and runs on windows 1:1 with uncore matching RAM speed. This is why I stick to 1.5:1 or 2:1 for boards that force the uncore-memclk ratio—since a 1:1 setup works well for stable high-frequency operation. They didn’t add special handling for this, so they defaulted to simpler settings. It’s still doable to run 200-220 bclk safely, unless you hit a bad board like my p6x58de. Adjusting IOH voltage might help, but that’s about it. By default I aim for at least 200 bclk and prefer around 210-225, which is why I choose 191 for 4.2ghz—because the sample I used doesn’t even reach a stable clock. 194 works for 3100 ddr3 on my i7 930 since the IMX tops out at that speed with samsung gdies.