PP Mguire
Build Guru
- Messages
- 32,592
- Location
- Fort Worth, Texas
Basically.
Bulldozer (V1 and V2) - The (partial) Scoop
- Thread starter 4P_Bulldozer
- Start date
- Status
It will be a tremendous success as soon as a real benchmark is developed. It just not a fair benchmark because, I'm assuming, AMD scores crappy.
AMD Will Not Endorse SYSmark 2012 Benchmark
AMD Will Not Endorse SYSmark 2012 Benchmark
Apokalipse
Golden Master
- Messages
- 14,559
- Location
- Melbourne, Australia
Sysmark does tend to favor Intel chips.
I know at least some versions use the Intel compiler, which deliberately creates slower code for non-Intel chips.
We'll have to see if newer versions of the Intel compiler have removed that functionality after the lawsuit was settled.
I know at least some versions use the Intel compiler, which deliberately creates slower code for non-Intel chips.
We'll have to see if newer versions of the Intel compiler have removed that functionality after the lawsuit was settled.
OP
OP
4P_Bulldozer
In Runtime
- Messages
- 205
- Location
- Earth
That is new News (today), they used to be on better terms: http://sites.amd.com/us/atwork/promo/sysmark/Pages/sysmark.aspx .
I guess you have no comment on the SB B2 Stepping "Cougar Point" chip set.
Perhaps you liked the Phenom 2 better (@ 7.02GHz) (Featuring Chew*):
gurusan
Golden Master
- Messages
- 6,562
- Location
- Portland, OR
Seeing as this is a Bulldozer thread I don't see how it's relevant. Not to mention the B2 stepping merely had "potential" SATA3 controller issues...nothing to do with the actual performance. However, yes....B2 stepping boards were a massive fail.
Yeah Chew's OCing is impressive...however I'm more into 24/7 daily clocks and real world performance than LN2 cooled systems running synthetic benches.
Sysmark is the most useless benchmarking software out there. AMD reps are not the only people who think it's a stupid benchmark. It's the epitome of synthetic benchmarks that has pretty much no relevance to everyday use.
Apokalipse
Golden Master
- Messages
- 14,559
- Location
- Melbourne, Australia
I think he was trying to make an analogy - that early chips can and often do have issues, and do not necessarily represent their true potential.
Yes, even Via and Nvidia think so.
AMD is playing Bulldozer close to their vests. I'm not exactly sure as to why other than they do not want anyone to steal their thunder. I do hope that the whole Bulldozer family does make a good showing in order to keep competition alive.
Apokalipse
Golden Master
- Messages
- 14,559
- Location
- Melbourne, Australia
1. it keeps Intel in the dark and therefore unable to determine how to respond (prices, product targeting), therefore giving AMD a competitive advantage.
2. AMD also has current chips to sell. Releasing all information about Bulldozer now would cause sales drop.
OP
OP
4P_Bulldozer
In Runtime
- Messages
- 205
- Location
- Earth
I was thinking that about the "C" and "E" Series, and that people would wait for the Llano, but those Chips are all sold out. If you need something that is ahead of the others today (the "C", "E", and "A" for graphics performance per Watt) then you buy now and maybe again in a few months or a year.
People who build SuperComputers are buying Bulldozers by the Pallet to replace their Opterons. There exists little question that the Chip is fairly great, if not "super". It is also great that they are being well tested before we (the latecomers) buy them.
When the general public gets them they will be unable to utilize all the power for some time to come (thus not needing to upgrade right away). Our 'upgrades' will come in the form of new Compilers and learning to use what we have (Parallel Programming).
The XK6 holds 768 Bulldozers, that is more than you will purchase 'San' so I don't think anyone is concerned about loosing your support. It the Chips did not work then there would be no purpose in upgrading last year's Computers with them (at great expense) or using them to build new ones. Take you feet out of your mouth so you can eat your words.
Sources:
gurusan
Golden Master
- Messages
- 6,562
- Location
- Portland, OR
Actually I am one of those people who build supercomputers.......I build rack servers at sun microsystems/oracle for a living. Guess what, we have the fastest database servers in the world and use multiple hexcore xeons in each node. When something faster and more power efficient comes out I'm sure we'll be implementing them
- Status
