• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

AMD thuban

B

Boostage

Limp Gawd
Joined
Sep 25, 2008
Messages
327
I can't find any recent news on this. every google search result is from Sept 09. Are there any updates on this chip?
 
there hasnt been any updates on it yet.. not sure why AMD is staying so quiet about the chip.. though they kept the phenom II's pretty quiet for a while too.. could be they are waiting for legitimate numbers on the i9's before they release any of their own numbers..
 
I didnt even think it was competing with the gulftown other than it having the same amount of cores. but the pII's have the same amount of cores as an i7 and they dont really compete. I hope AMD release some data soon.
 
Boostage said:
I didnt even think it was competing with the gulftown other than it having the same amount of cores. but the pII's have the same amount of cores as an i7 and they dont really compete. I hope AMD release some data soon.
Click to expand...


it probably wont compete with it.. but it could also determine what the final clocks will be on the high end thuban chips.. but at the same time they could also be running into problems with the TDP on the chips.. since they need to stay under 140w to work on the current generation of amd motherboards..
 
All of their 6-core chips are being put to use in server applications right now. Since the Deneb and it's descendants are selling quite well for AMD, they probably don't feel a need to rush them out. The Istanbul will bring a much larger profit margin for them, so it's what they'll focus on.
 
Recent news just got posted in the month-old "975" thread. There isn't a whole lot of new information, but a tiny bit apparently got given out with their launch of Phenom II/Athlon II refresh parts.
 
Every time you add cores you still have to work within the thermal envelop that motherboards can support. 2.8 at 140 for six cores seems appropriate. The max we can get right now is 3.4 at 125w. Add two more cores and i'm glad we're even getting 2.8 at 140. It's an engineering issue and in all reality most people on here seem to be gamers and until games support even quads the six cores will most be for people doing video trans-coding etc. They care about threads more then individual core top speed. That's why the Athlon 620 for 99 bucks is so hot right now.
 
OFaceSIG said:
Every time you add cores you still have to work within the thermal envelop that motherboards can support. 2.8 at 140 for six cores seems appropriate. The max we can get right now is 3.4 at 125w. Add two more cores and i'm glad we're even getting 2.8 at 140. It's an engineering issue and in all reality most people on here seem to be gamers and until games support even quads the six cores will most be for people doing video trans-coding etc. They care about threads more then individual core top speed. That's why the Athlon 620 for 99 bucks is so hot right now.
Click to expand...


exactly..

R3MF said:
fastest thuban only 2.8GHz at 140W?
Click to expand...

do the math.. 3.4ghz x 4 = 13.6ghz / 2.8 x 6 = 16.8ghz so you gain 3.2ghz for multi-threaded applications while staying within a 140w limit(and dont try to claim you can do that by overclocking a 965 to 4ghz because at 4ghz the TDP is well above 140w).. doesnt mean it wont be overclockable.. just means that amd has stuck to its set limit which is 140w.. a good example is the ATI HD5970.. they had a set limit of 300w for the card.. yet right from the start they were telling and providing software to overclock the card..
 
Last edited:
drescherjm said:
This assumes 100% efficiency in scaling which is never achieved.
Click to expand...

It is if you are using the chips for things that take place on one core. For example if you were using these to crunch, you get to do 6 tasks at once at 2.8Ghz instead of 4 at 3.4Ghz. Even in things like H264 encoding 6 cores at 2.8Ghz is going to steamroll 4 cores at 3.4Ghz. I don't think he was intending to give the math as an exact performance comparison, just a simple work/watt indication. You get more potential work done with the same amount of power used. Throw on the fact these will probably hit into the 3Ghz territory very easily on stock volts (if current quads can be used for an example) and you have some potent chips for those with AM2+/AM3 systems looking to upgrade, or those wanting nice performance on a budget.
 
Based on the diagram, these seem like exact copy's of the Opteron 24xx series, everything is exactly the same but one thing, a 200mhz increase in speed over the top Opteron model. I don't believe this product will be revolutionary in any way.
 
The reduction in price over the Opterons will be nice. This is basically a workstation processor available on the same motherboard as the desktop line.

However for over 90% of users this is not the chip to buy, it will be slower in most applications than the 955 and cost twice as much.
 
I just kinda wish they would shrink their process to 32nm would be ideal on this, because at 45nm at 2.8ghz at 125W isn't really ideal, but we'll see how it performs.
 
2.8GHz would not be bad if it had a 20% (or greater) IPC advantage over current Phenom2s.
 
In 2009, AMD Opteron 2435 X6 2.6GHz (Istanbul) was released which is based on Deneb architecture (Thuban is also based on Deneb, just with a lower TDP). If you're interested in how well this CPU stacks up against quad-core P2 X4 965 3.4GHz, P2 X4 910 2.6GHz and i7 920, check this link: http://www.xbitlabs.com/articles/cpu/print/amd-istanbul.html . After doing some simple maths we get that 2.8GHz X6 Thuban will perform like ~3,4GHz quad core Nehalem based CPU.
 
I do not see any way the the 2.8GHz thuban would compare to a 3.4GHz i7. Those benchmarks certainly do not show that. At 2.8GHz on those benchmarks the thuban can possibly be in the same performance class as the i7 920 which will be discontinued well before the launch of the thuban.
 
Last edited:
drescherjm said:
I do not see any way the the 2.8GHz thuban would compare to a 3.4GHz i7. Those benchmarks certainly do not show that. At 2.8GHz on those benchmarks the 2.8GHz thuban can possibly be in the same performance class as the i7 920.
Click to expand...

No, they don't. Keep in mind, that one Nehalem core, where SSE4.1, SSE4.2 instructions are used, is ~30% faster than Phenom II at the same clock speed (taken as an average from benchmarks like video encoding..) 6 cores is 50% of increase over 4 cores, so 50 - 30 = 20%. 20% of 2800MHz is 560MHz so, 2800 + 560 = 3360MHz.
 
But there are 8 cores on the i7 with the 4 virtual cores. This accounts for something. Also you can never assume 100% scaling.
 
drescherjm said:
But there are 8 cores on the i7 with the 4 virtual cores. This accounts for something. Also you can never assume 100% scaling.
Click to expand...

in heavily threaded apps, the scaling is incredibly close to 100% for extra cores,

http://it.anandtech.com/IT/showdoc.aspx?i=3571&p=9

in some cases, it's within the margin of error 1-2% not to mention that the quad is 2.7ghz in that and the 6core is 2.6ghz
 
That can only be achieved on very specific workloads that do not require a large memory /cache bandwidth. So there is no resource contention between the cores.
 
It looks like the Thuban will compete with i7 on some levels. Hopefully it overclocks as well as the x4s. If I can watercool a Thuban and get 3.6 Ghz out of it stable with F@H on my current mobo I'm going for it.
 
drescherjm said:
That can only be achieved on very specific workloads that do not require a large memory /cache bandwidth. So there is no resource contention between the cores.
Click to expand...

I was under the impression that the HT & Cache setup on the PII chips is more then sufficient to feed 6 cores =)
 
You must log in or register to reply here.
Back
Top