Memory Scaling on Core i7 - Is DDR3-1066 Really the Best Choice?
by Gary Key on June 24, 2009 9:00 AM EST- Posted in
- Memory
And we begin, with a graph:
The graph above represents the cost, from Newegg.com, of 11 different 6GB DDR3 memory kits (1066 C7/C5 are the same kit). The only variables are the manufacturer and speed of the DDR3 memory included in the kit.
The least expensive DDR3-1066 6GB kit we purchased sells for $80, the most expensive 6GB kit? $289. That’s over a $200 difference; that and some pocket change is enough to pay for a sweet new video card, a nice 22” monitor, or even a iPhone 3GS.
All of that extra money is going somewhere: frequency and latency. The lowest end kit has a data rate of 1066MHz and a CAS latency of 7 cycles. The most expensive kit has a 1866MHz data rate at the same CAS latency; that’s a 75% increase in data rate.
A 10% increase in CPU speed rarely yields more than a 4 or 5% increase in performance, but what about a 75% increase in memory speed? We don’t have a single page on AnandTech to point you to that would answer that question. At least we didn’t, until today.
A Brief History of DDR3
We first met DDR3 alongside Intel’s P35 chipset. It’s performance at the time was at best equal to or usually worse than DDR2 while carrying a significant price premium. It wasn’t until the release of the Intel X48 and NVIDIA 790i chipsets that Socket 775 users could even see an advantage to using DDR3 and even then, it was primarily for benchmarking contests - hooray, because we all know how important those are.
There were some tangible advantages to DDR3 from the start, the biggest being it’s lower operating voltage. DDR2 memory required 1.8V while DDR3 could run at 1.5V, this made DDR3 particularly attractive for notebooks but on the desktop the advantage was sort of abused.
In order to take advantage of DDR3’s higher memory speed, benchmarking enthusiasts often had to use DDR3-1866/2000 kits that required voltages in the 1.8V~2.0V range to reach these clock speeds. Performance improvements in certain benchmarks were available through brute force use of voltages and ICs that allowed high memory speeds at decent latencies. In the end, improvements in actual applications were just not worth the cost or trouble of using DDR3 compared to DDR2.
Intel made no secret of its plans to move the vast majority of their processor lineups to DDR3 memory a few years ago. They truly wanted/desired that DDR3 would be the standard memory of choice by the time P45/X48 launched, but DDR3’s limited availability, middling performance, and a price premium that was truly prohibitive for the mainstream crowd, let alone the enthusiast market space made this desire impossible. As such, this left users with memory controllers not truly optimized for either memory technology that resulted in performance not significantly different from earlier Core 2 supporting S775 chipsets.
AMD recently released the AM3 platform and their AM3 based Phenom II processors support both DDR2 and DDR3 allowing users to either upgrade to AM3 boards or keep using their DDR2 based AM2+ boards, a simple if not elegant solution. With the recent release of new boards, BIOS optimizations, and near cost parity with DDR2, we now think DDR3 is a viable option for AMD users.
That said, DDR3 started coming into its prime last fall as the release of second and third generation DDR3 devices along with Intel’s Core i7. Core i7 and the X58 platform introduced Intel’s first on-die memory controller. Aside from other architectural improvements, Core i7 supports three DDR3 memory channels compared to the standard two channel setup, yielding some impressive bandwidth numbers.
There was one problem with this new design, a real fly in the ointment actually for the memory suppliers. Current JDEC specifications list 1.50V as the official voltage specification for DDR3 with a move to 1.35V in the near future and eventually to 1.20V. However, most of the performance oriented DDR3 modules released for the Core 2 platforms generally operated at 1.8V~2.0V in order to hit high speeds with decent latencies. In essence, the memory kits were already overclocked to hit clock speeds that made DDR3 a performance consideration on S775. The memory controllers on the Core 2 based Northbridge products could easily handle these voltages and surprisingly enough, so could the ICs.
Well, Intel officially released their recommended memory voltages for the Core i7 several months before launch with 1.50V as the recommended base voltage and 1.65V as the suggested maximum along with 1.35V for VTT (QPI) maximum. This left the memory suppliers in a bind as true high-speed low-voltage ICs were not going to appear until after the Core i7 launch. Intel extended official support to DDR3-800/1066 speeds only, although most current X58 motherboards support speeds up to DDR3-2133 or higher.
At the Core i7 launch, the market ended up with a bevy of DDR3-1066/1333 low-voltage kits with a sprinkling of higher speed DDR3-1600 and DDR3-1866 kits based on highly binned ICs that met the 1.65V recommendation. One other twist was that the memory suppliers had to package low-voltage higher-speed 3-DIMM kits instead of the normal 2-DIMM configuration in order to satisfy triple channel owners.
Fast forward to today and we see the wide availability of high-speed low-voltage DDR3 products with more choices coming on a daily basis. In fact, the availability of DDR3 products almost matches that of DDR2 with similar price points. By the end of this year, DDR3 products will outnumber DDR2 offerings although it will be a couple of years before DDR3 usage overtakes DDR2.
47 Comments
View All Comments
sonci - Thursday, June 25, 2009 - link
The title was enough,and yours is too..
goinginstyle - Wednesday, June 24, 2009 - link
I have no idea where you have been hiding Gary but it is great to see you back. I look forward to the AMD article and the individual module tests. If they are half as good as this, I will be one happy person. Any chance or comments on Virtualization benches under WIn7 with XP?duploxxx - Wednesday, June 24, 2009 - link
very nice article with real world applications, now can you pls do the same with amount of dimms used and finally kill this marketing hype?vailr - Wednesday, June 24, 2009 - link
Re:"Current JDEC specifications list 1.50V as the official voltage specification for DDR3 with a move to 1.35V in the near future and eventually to 1.20V."
How does DDR5 (already being used on some video cards) voltages compare with those numbers?
Any predictions on when DDR5 memory modules for main boards might be introduced?
Are lower latency DDR3 modules "in the works"?
Maybe such a thing as:
1333 MHz @4-4-4-12 & 1.20v, for example?
Also: Intel promotes the benefits of using Hafnium when producing their CPU's. Would Hafnium also benefit memory chips?
GourdFreeMan - Thursday, June 25, 2009 - link
The video cards you are thinking of use GDDR, not DDR. You will never find GDDR packaged on DIMMs for motherboard memory slots, but rather only as individual chips for graphics boards and consoles. Though they share some technologies, you would need a different memory controller to make use of it. Typically GDDR5 runs from 1V to 1.4V (the GDDR5 on the stock Radeon HD 4770 runs at 1.263V for example). Hafnium dioxide has already been used by some manufacturers as the dielectric material in DRAM capacitors, and I know NEC, at least, has already used Hafnium in transistor gates for embedded DRAM. As for higher speed and lower latency RAM, that is pretty much always in the works...Zorlac - Wednesday, June 24, 2009 - link
I have thought this all along, but was too lazy to prove it. Thanks Anand!!! :)Any idea when we will start seeing 4GB DIMMs for running 3x4GB kits?
Gary Key - Wednesday, June 24, 2009 - link
We just finished testing a 12GB 1600 C9 kit from Kingston, but the street price is $1400. I also have a 24GB kit from Corsair, but I will not even mention the cost on that one. We should see affordable (compared to buying 6x2GB) 12GB kits later this year in the 1333 C8 variety without ECC. I do have a 12GB ECC Kingston 1066 C8 kit arriving shortly, retail is about $320, to test in some workstation products.DXRick - Wednesday, June 24, 2009 - link
It looks like the sweet spot is 1600 C8 at around $100 for a 6BG kit.vol7ron - Wednesday, June 24, 2009 - link
agreed --- those saying 1333 really are saving the buckI would still like to see SuperPi tests, since they are a mathematical approach to memory performance.
Gary Key - Wednesday, June 24, 2009 - link
Super Pi results will come in Raja's 2000 shootout. However, most of the Super Pi program runs in cache and when it goes outside of it,it only hits a couple of ranks in short bursts (why bandwidth matters) and as a result a proper loading of the entire memory subsystem is not really tested thoroughly. That said, we will have a complete subsection on Super Pi for the overclocking crowd. :)