HyperTransport and Opteron/Athlon64 Overclocking
The first question many will have about our efforts to look at how Athlon64 will perform is how we can possibly compare an overclocked Opteron to a chip that is not overclocked. In the case of the Opteron, the comparison is more accurate than you might first think.In normal setups (e.g. Athlon/P4), the CPU gets its clock from the FSB clock and multiplies it by the “clock multiplier” to determine how fast its internal clock should be. With a 16x multiplier, when the external clock ticks once, the CPU ticks 16 times. However, with the Athlon 64/Opteron, there is no FSB, so the CPU must get its clock from somewhere. It doesn't produce it internally; instead, it derives it from the native HT (HyperTransport) frequency, which is 200MHz, but because of the bus' nature, it runs at an effective 800MHz.
So, for our 1.8GHz Opteron 144, the multiplier is 9x, which is why raising the HT frequency to 222MHz increases the clock speed to around 2GHz. But we are increasing the HyperTransport clock in our overclocking, and not a FSB clock, which does not exist on Opteron/Athlon64. In real terms, this means our CPU overclocking has a significant impact on Performance, but it is unlikely that our increase in memory speed will have nearly as much impact on performance. Since we are nowhere near saturating the Hypertransport bus at 200 (effective 800), increasing HyperTransport to 222 (888) will not likely have much, if any, impact on overall performance. Our performance improvements, with Opteron/Athlon64, are mainly coming from increase in CPU clock — much more so than on the Pentium 4 or Athlon architectures.
Obviously, the PCI bus operates at a different frequency than the HT bus than the CPU, but they all operate based on multiples of each other, and are all derived from the HyperTransport clock.
Facebook
Twitter
RSS
79 Comments
View All Comments
Anonymous User - Friday, September 5, 2003 - link
Hey Wes, I've got some minor nitpicking to do. The graphs look great, but there's no unit labels on them. I know what the labels should be, but when you're presenting data you should always be sure to include labels for everything on the plot. Otherwise, you run into the possibility of misrepresenting data. Please make sure that all the graphs on future previews/reviews have their units labeled and perhaps even "bigger is better/shorter is faster"-like comments. Just addressing a pet peeve of mine. Aside from that, great article! Thanks.Anonymous User - Friday, September 5, 2003 - link
This is not exactly related but for future motherboards/chipsets using this upcoming processor please use a RD2 PC Geiger (http://www.ioss.com.tw/web/English/RD2PCGeiger.htm... that you may provide us with information about the PCI bus frequency - important for overclocking.Anonymous User - Friday, September 5, 2003 - link
#25. Ive tried both intel and AMD solutions for my personal machines at home. 10 times out of 10, AMd's have little quirks with them. the ride the bleeding edge.The company is losing money and cant continue to dump into R&D much longer, while INTEL's stock doubled in price in the last 4 months and is still rated a five star buy buy buy. Markets dont lie. The money is betting against AMD and in the fast moving techincal processor market where R&D is the most important aspect in the business model, AMD is losing. They just dont have the cash or assets to compete. The only thing they are good for is controlling The price Intel charges.
Anonymous User - Friday, September 5, 2003 - link
hey #25 amd cant even get a cpu to the 3 ghz range. They had to create a virtual 3200+. the reason is an athalon runnig at 3 ghz is a FIRE HAZARD.Anonymous User - Friday, September 5, 2003 - link
#24, that's funny! I didn't think that Intel fanboys still read AMD articles because in the end Intel is always better right?
So why waste your time reading the article if you already know what is best? Do you feel threatened that you won't have the best CPU anymore?
Go and cry to momy.
Personally, I found this article very educational because I had my mind set on buying an Intel 2.4C CPU and overclocking to 1000 FSB with DDR 500 but now I'll wait a couple more weeks to see how the Athlon64 turns out and how overclockable it is.
I'd also like to thank AnandTech for increasing the number of articles produced in the last week. If this continues then I'll become a regular visitor.
Anonymous User - Friday, September 5, 2003 - link
Face it. Intel makes the far superior Processor. That have the money to dump into R&D. This is a sign of the end for AMD. The farther they seperate themselves from the maintream, the more they lose.Who cares if they have some loyal geeks they worship them.
The real money is made by INTEL in the (Say it with me) BUSINESS market.
Their is a reason why dell wont touch an AMD
Its called UNRELIABLE
Anonymous User - Friday, September 5, 2003 - link
"Apparantly" socket 940 uses Registered dimms only, the boards do have overclocking options and DO overclock quite well.If you wait till QTR1 2004 you will be able to find socket 939 boards that do everything the 940 boards do, but can do this with Unbuffered Dimms also.
So if you want to upgrade but don't want to swap your standard PC3200 wait till early 2004 before you buy.
MS - Friday, September 5, 2003 - link
Wes,Which settings are you using in the GunMetal Benchmarks? I cannot replicate your results there and I am wondering whether I am doing something wrong ..
TIA
Anonymous User - Friday, September 5, 2003 - link
#20, 99% of gamers don't care about workstation benchmarks either :)Anonymous User - Friday, September 5, 2003 - link
#19, no one runs Quake 3 in Linux. 99% of gamers use Windows, it would be a complete waste of time to do Linux Quake 3 benchmarks.