Conroe is 4 months away, not 6. And personally i'm not gonna get a Conroe no matter how good they are. If i get anything itll be a cheap X2 or Presler.
dug777
Lifer
- Oct 13, 2004
- 24,778
- 4
- 0
Those PResler prices will obviously drop before then, so they will be muc hcheaper, though i'd bet that the slow ones will get the EOL, so they wont ever get too cheap. Also, both Conroe and Presler are 65nm, so that don't help the price of either.
Some quick facts:
1. 2.66ghz isn't the top rated part. It will be a slightly above midrange offering.
2. There was no fancy cooling on the 2.66ghz part, because...
3. The 2.66ghz was running below default voltage, hence...
4. Good merom bins can run at 3.33ghz.
But if anyone wants to cry shenanigans on the merom numbers (could it be the same people who cry foul over AM2 previews... LOL), knock yourself out.
Zebo
Elite Member
- Jul 29, 2001
- 39,398
- 19
- 81
We'll see.
Oh and I never said Conroe won't beat X2 - and I've already said AM2 will be no gain at best. So no need to be smug about it.
What I have said is Intel no doubt showed tasks it would "wow" best with. AMD will be much closer than 20% in some benchmarks.
Oh and I never said Conroe won't beat X2 - and I've already said AM2 will be no gain at best. So no need to be smug about it.
What I have said is Intel no doubt showed tasks it would "wow" best with. AMD will be much closer than 20% in some benchmarks.
yeah, 'cause i calculate millions of Pi terms a day in my everyday life...
its synthetic alright, just like 3DMark et al...
But that doesn't mean synthetic benchmarks have no place here. Personally I don't know of anyway to run a timedemo in CAD or Excel, so you need office type benchmark tools. But SuperPi is pretty far into the synthetic range for me, its not even trying to emulate any real life programs.
its synthetic alright, just like 3DMark et al...
But that doesn't mean synthetic benchmarks have no place here. Personally I don't know of anyway to run a timedemo in CAD or Excel, so you need office type benchmark tools. But SuperPi is pretty far into the synthetic range for me, its not even trying to emulate any real life programs.
Synthetic benchmarks emulate workloads which are not specific enough for explicit modeling, so they actually model "real work" as defined by the average user much better than anything else, since most people use web/productivity.
On the other hand, superpi uses the gauss-legendre algorithm, but it's closed source so the guts cannot be easily examined on a whim. But it can be assumed it is a heavy exercise of both int and float operations since the method contains arithmetic which requires both functional units. Kinda like games, actually. So while it is not typical work for an average user, I'd argue it covers the scientific/gaming side of usage models.
On the other hand, superpi uses the gauss-legendre algorithm, but it's closed source so the guts cannot be easily examined on a whim. But it can be assumed it is a heavy exercise of both int and float operations since the method contains arithmetic which requires both functional units. Kinda like games, actually. So while it is not typical work for an average user, I'd argue it covers the scientific/gaming side of usage models.
RichUK
Lifer
- Feb 14, 2005
- 10,341
- 678
- 126
Synthetic benchmarks are only good for benchmarking or comparing the same model or similar model processors. Its no good for cross examining two completely different platforms, as the benchmark can be bias, and will not indicate real world performance. That is the point of view I am coming from.
he's probably right . the 2.66 desktop chips have to run at a 65 watt power ceiling, but the woodcrests have a max 80 watts.
they did have a 3.0 ghz woodcrest there, and woodcrest has 1333 bus. im sure the EE will be like 3.33 ghz
Originally posted by: Zebo
If conroe will clock so high and be so effceint why is only 2.66Ghz it's top rated and they go way down to 1.86 from there?
I don't know maybe 2.66 is enough to beat 2.8 X2 like intel claims and that's all they want, to beat, not to cream. Maybe they really do have this 3.3Ghz XE conroe rumored.Or maybe the conroe simply won't ramp very well due to short pipeline. Or maybe it starts leaking like crazy after 2.66 sucking loads of power!
Wait and see.
3.33ghz on 65nm is on the roadmap.
An accurate definition of Synthetic bechmarks is:
"Synthetic benchmarks, by definition, aren?t real programs, and their code doesn?t compute anything anyone would necessarily ever want to compute. As such, the results they yield may or may not be useful in predicting the performance of real programs. There are a whole host of problems with synthetic benchmarks; the topic is worthy of an article in and of itself. Hennessy and Patterson consider these benchmarks the least useful of all the types discussed."
http://arstechnica.com/cpu/2q99/benchmarking-2.html
"Their code doesn?t compute anything anyone would necessarily ever want to compute", Ok superpi fulfill that condition, Who the hell needs to compute millions of decimal digits of pi?
"the results they yield may or may not be useful in predicting the performance of real programs" True, that's why superpi is not usefull predicting real world performance as we see in the bechmarks.
CONCLUSION: Superpi IS Synthetic.
per your definition
e.g., you buy F.E.A.R just to watch the demo ?
come on, benchmark is benchmark. Syn or not, it is never your real life computing experience.
when you do benchmark, syn or not, it always tell you something, except the case where the hardware is particularly "optimized".
all benchmarks are synthetic
e.g., you buy F.E.A.R just to watch the demo ?
come on, benchmark is benchmark. Syn or not, it is never your real life computing experience.
when you do benchmark, syn or not, it always tell you something, except the case where the hardware is particularly "optimized".
That is a crap definition. Most benchmarks don't compute anything useful. So what's a real program? Nobody wants to try to create a benchmark set based on office. If a synthetic benchmark is able to emulate tasks typical to such workloads, then it is a good indicator of performance of a machine on aforementioned "real programs". Naturally academics consider synthetics useless. They only care about high performance suites... tests like superpi, for example. LOL.
One can argue that a CPU that is good at torture tests will naturally perform well at well-crafted synthetics... which might be true. But then again, the whole computing platform has a bigger effect on productivity synthetics than just the CPU, but that is a whole new argument.
Oh fyi, the superpi program has an explicit computation, so by definition it's a real program.
Some people do, and it is not too different from the typical scientific. Since when was it your perogative to dismiss such workloads?
I'm too busy to research, but I suspect superpi has a decent correlation with games, a benchmark which can also be defined as synthetic by some, but real to others. Are games synthetic? Whatever.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 22K
-
News NVIDIA and Intel to Develop AI Infrastructure and Personal Computing Products- Started by poke01
- Replies: 374
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
Facebook
Twitter