Phenom B3 is right around the corner...
by Gary Key on February 9, 2008 1:00 AM EST- Posted in
- Gary's First Looks
The bouncing ball release schedule for the long awaited Phenom B3 revision that fixes the Translation Lookaside Buffer (TLB) errata has taken a favorable bounce according to our sources. Advanced Micro Devices (AMD) plans on releasing three versions into the retail market in the early April time frame. Fortunately for consumers, it appears that AMD will change the product designations in order to differentiate the B3 series from the current B2 release.
Unfortunately, we will not see a significant increase in bin speeds as the new models are coming in at 2.2GHz (9550), 2.3GHz (9650), and the 2.4GHz (9750). It also appears the current pricing structure will remain somewhat in tact although a final decision has not been made by AMD. Also, there will be a BE edition, but once again whether that will occur with the 2.3GHz (9650) or 2.4GHz (9750) is not finalized.
We originally expected the 2.4GHz (9700) and 2.6GHz (9900) parts to be released in late Q1, but after the TLB errata fiasco, those products were basically dropped in favor of getting the B3 series ready. Our sources indicate the Phenom 9950 being launched in late Q2, possibly Q3 at this time with a clock speed of 2.6GHz. This places AMD once again in an awkward position of not being able to respond in a timely manner to Intel's new Penryn based Quad-Cores that will be available shortly at speeds anywhere from 2.5GHz (Q9300) to 3.2GHz (QX9770). The current 3.0GHz (QX9650) being available for a few weeks now.
The good news is the TLB errata fix in the B3 will not incur a performance penalty when compared to the current BIOS enabled solution. A solution that so far causes performance penalties up to 18% in certain applications and BIOS configurations. While the current beta version of AMD's AOD utility allows you to turn off the TLB fix, we have found performance still suffers slightly. This also holds true with current BIOS releases that offer an enable/disable function compared to early release BIOS spins that ignored the TLB problem. We hope to have a B3 based Phenom shortly for testing to verify AMD's performance claims.
Facebook
Twitter
RSS
15 Comments
View All Comments
initialised - Tuesday, February 12, 2008 - link
http://www.nordichardware.com/news,7328.html">http://www.nordichardware.com/news,7328.html"If you feed the data into the validator you get an incorrect checksum and the CPU-Z is horribly out of date. Consider it with a truckloads and truckloads of salt."
initialised - Tuesday, February 12, 2008 - link
Nehalem's 8s superPI is impressive. SuperPI is a single core test for Phenom and Core2 so I suspect that Nehalem may be parallelising it to half the time.With any luck I'll have some B3 data by the end of the week (NDAs permitting). I'll try and get SuperPI 1M data with good RAM.
JumpingJack - Monday, February 11, 2008 - link
"The current 3.0GHz (QX9650) being available for a few MONTHS now." -- corrected.(Quote button not working for me)....Anand actually did the pre-launch review on Oct 29th, the processor officially launched the 14th of november, and was seen on the shelves about a week later as I recall:
http://www.anandtech.com/cpuchipsets/intel/showdoc...">http://www.anandtech.com/cpuchipsets/intel/showdoc...
http://www.hothardware.com/articles/Intel_Official...">http://www.hothardware.com/articles/Int...fficiall...
http://www.dailytech.com/Intel+Officially+Launches...">http://www.dailytech.com/Intel+Official...5nm+Penr...
"While the current beta version of AMD's AOD utility allows you to turn off the TLB fix, we have found performance still suffers slightly. " How can you tell that? Do you have B3, how do you know B3 did not implement something that slightly improved over B2? This is an odd statement, not backable by facts.
JarredWalton - Tuesday, February 12, 2008 - link
I believe what he's saying is that with B2 CPUs, even with the BIOS fix turned off, performance is lower than it was prior to the introduction of the BIOS fix. So the BIOS got de-tuned a bit in the process, no matter whether you enable or disable the TLB fix. B3, so far as I know, is not in our possession yet. (Though Anand might have one for all I know....)Bozo Galora - Monday, February 11, 2008 - link
Well gee, maybe when they get around to selling this thing or its offspring to the public at large, it will be able to compete with Nehalem?Oh golly, an ES STOCK 2.66 non overclocked Nehsalem does an 8 second 1M super pi.
http://infomars.fr/forum/index.php?showtopic=1524">http://infomars.fr/forum/index.php?showtopic=1524
Hmmmmmmmmm.
Bozo Galora - Monday, February 11, 2008 - link
Then again, maybe it wonthttp://www.xtremesystems.org/forums/showthread.php...">http://www.xtremesystems.org/forums/showthread.php...
diablosinc - Monday, February 11, 2008 - link
what was the point of that link? the last post in that thread was july '06!!!Pale Rider - Monday, February 11, 2008 - link
Are they kidding us? We have already been waiting years for this chip. This is getting as bad as Duke Nukem Forever... wait any longer and everyone will just go back to Quake...JumpingJack - Monday, February 11, 2008 - link
Well, the chip actually launched Nov 19th, and the TLB errata is way overblown. Having assembled a 9600 BE, I have yet to run into anything I would call stability issues -- I run without the 'fix' in place, disabled by BIOS. AMD's overclocking utility is doing more with the green, yellow, red than just disabling Errata 298, what all it is doing I am still trying to figure out... but it appears to be disabling other workarounds for errata other than the TLB bug.fitten - Sunday, February 10, 2008 - link
Or "intact"?As an above poster commented, has AMD indicated that B3 will fix the AMD K10 "core2" issue that's currently being seen now?