cancel
Showing results for 
Search instead for 
Did you mean: 

Memory Bandwidth Booster - What does it do?

Stu
Level 10
Hi guys,
Do any of you have any luck with this settings? Any performance gains?

Asus Maximus IV Extreme Z
i7 3770K @ 4.4ghz
AMD 7970 HD x3 in TriFire @ 1200/1800.
2x 240GHZ OCZ Vertex 3 SSD & 1x 512Gb Vertex 4
16GB 2133mhz Patriot DDR3
Ennermax Max Revo 1500w PSU
Huge water loop with an internal EK 340mm & external Supernova 1260mm
Windows 7 64 bit.

Aquaero Fan control & monitoring
13,098 Views
10 REPLIES 10

Retired
Not applicable
No... Its crap

Spathi
Level 9
It tweaks the timings under the main timings for benchmarking for very specific high end memory (that ASUS have not specified). It might be an obsolete feature as I don't think Elpida Hyper works well in newer PCs and Elpida went into administration. Maybe there is some other brand that has a high end module that would work with it or maybe it is just a bug fix for some ICs. Anyway I think it is just for (fun if it works) benchmarking only on 1x or 2x smaller than 4G sticks and not to be used for daily computing.

Memory on Sandy is already bottlenecked by the CPU @ 1600MHz as the cache on Sandy (although good/better than all mass consumer CPU's b4 it) is the limiting factor. Think of memory as the hard disk for the CPU and the CPU with its memory (Lx caches). Ivy and its slightly optimized logic and slightly larger cache might be faster with 1866 and above, but I have not seen the benches yet. Tweaktown is about to do one. They compared Sandy and Ivy with 1600MHz memory .. which is silly, but I think they are doing overclocked next.

Anyway it is reduced latencies / (and maybe combined with some increased latencies) with sub-timings, not increased MHz or anything.

HiVizMan
Level 40
My dear cave dweller friend has hit the head of that nail rather succinctly - if somewhat roughly. 😄

Do not bother with it.
To help us help you - please provide as much information about your system and the problem as possible.

I didn't know that the limiting factor was the CPU @ 1600MHz. So high end memory faster than 1600 is somewhat of a waste of $?

I am using 2 x 4Gb of Corsair Vengeance PC3-10700 (667MHz) - [DDR3-1333] and was considering an upgrade to the Dominator style.
ASUS Rampage V Extreme BIOS 4101 | i7-6950X | Thermaltake Core X9 | G.Skill F4-2800C16Q-32GRK | Cooler Master Nepton 280L | Dual Samsung 850EVO 500GB SSD | PSU: DARK POWER PRO 11 1000W | 3TB & 4TB HDD | NVIDIA GTX Titan X | ASUS 24x DVD±RW Drive | Win10 Pro

HiVizMan
Level 40
Not sure I would agree with that statement about memory being bottlenecked on the SB CPU however the OP is about the Mem Bandwidth Booster so I will stay on topic.

I run 2240Mhz on my memory (true my cpu is not at stock) and most certainly in certain benchmarks that use or need memory bandwidth I get real and measurable gains.
To help us help you - please provide as much information about your system and the problem as possible.

HiVizMan wrote:
Not sure I would agree with that statement about memory being bottlenecked on the SB CPU however the OP is about the Mem Bandwidth Booster so I will stay on topic.

I run 2240Mhz on my memory (true my cpu is not at stock) and most certainly in certain benchmarks that use or need memory bandwidth I get real and measurable gains.


What model # are sticks are you using ?

Spathi
Level 9
http://www.bit-tech.net/hardware/memory/2011/01/11/the-best-memory-for-sandy-bridge/1

The cache stack on sand it per core. On Ivy it is per hyperthread and there is more of it. Anyway I will be interested to see some results for Ivy. Sandy-E also has more cache so gets a lot of performance from that.

The only way to show a difference running an executable might be to use large instructions, which Sandy and below slow down on... http://www.agner.org/optimize/blog/read.php?i=142

Stu
Level 10
Thanks guys, benching my ass off this weekend and don't want to waste time testing stuff that doesn't work.

Bought myself some new Sammy 1600mhz 30nm ram and its so much faster than my Corsair 1800ram its untrue, well pleased with the performance gains out of it now its running at 2133 9/10/10/27/1 - 1.5v.


Appreciate your help always. 🙂

Asus Maximus IV Extreme Z
i7 3770K @ 4.4ghz
AMD 7970 HD x3 in TriFire @ 1200/1800.
2x 240GHZ OCZ Vertex 3 SSD & 1x 512Gb Vertex 4
16GB 2133mhz Patriot DDR3
Ennermax Max Revo 1500w PSU
Huge water loop with an internal EK 340mm & external Supernova 1260mm
Windows 7 64 bit.

Aquaero Fan control & monitoring

Spathi
Level 9
Well maybe I am wrong, but you could perform tests like in in bi-tech with OC'ed CPU's and see what happens. I am assuming that something in the CPU will be a limiting factor. bit OT though, sry

ahh ANAND did it, nvm

http://www.anandtech.com/show/4503/sandy-bridge-memory-scaling-choosing-the-best-ddr3/7