Nehalem-EP......BLOOMFIELD

[gallag]

Why would you say that. I am not a member of AMDZone, but i have frequented the site. A few of their poster are far more intelligent than any one here. (no offence intended). Some would refer this site to being far to Intel biased- No big deal though.

Kindest Regards

The thing is, This site is pro Intel. Intel have the performance crown so this site will swing to Intel. Just look at the number of people watching the AMD section compared to the Intel section. Now I don't know if you are a fanboy or not, If you are you will not understand this, If not you will. When AMD was murdering Intel In the p-4d vs x2 days the ratio of AMD and Intel viewers was the exact opposite of today, This alone should be compelling evidence that around here most people are performance biased. If AMD came out with k11 tomorrow and it kicked Intel's ass around here would swing to AMD again.

Remember, These posters on AMD zone that are smarter than most of us folks are adults that are emotionally attached to A large company to the point that they will attack you if you happen to mention any thing bad about said company. How can a fanboy be smart?

[JCornell]

CINEBENCH R10 continue


...

[Shintai]

35C idle, 40C load..nice

[JCornell]

RightMark Memory Analyzer v3.80


...

[JCornell]

35C idle, 40C load..nice

Still doubt of its accuracy though

[Gautam]

There are many advantages that I have read as to why CMOS is preferred over just NMOS, part in due to what you rhetorical question above ... while I understand the physics, I am not completely well versed on the actual design side ... constructing an inverter in CMOS has an advantage that it consumes much less power overall. (NOTE: I had to cheat to make sure I got this right ... a good text on this subject is Solid State Electronic Devices by B. Streetman and S. Banerjee)

Jack

Simply put, in the example of the inverter, using an NMOS transistor for pull up wouldn't bring the output to the voltage of the power rail. The NMOS would act as a resistor, drawing and wasting power. The PMOS on the other hand would bring it up all the way (effectively a short) resulting in no static power dissipation. That's the big advantage of CMOS, not only that it usually ends up being one of the fastest options and the easiest to work with.

Pass transistor logic is supposedly a popular scheme involving only the usage of NMOS transistors...but all of my practical experience (in a simulator) with it has shown me that it's a nuisance.

[Shintai]

Still doubt of its accuracy though

If you touch the heatsink under load, how it feel? Unreasonable with the showed temps?

[JCornell]

If you touch the heatsink under load, how it feel? Unreasonable with the showed temps?

I think this one way more accurate , furthermore the software is not ready yet ...


...

[cpuz]

> JCornell
Thanks a lot for sharing these information

Do you see something on the mainboard that looks like a clock generator ?
(it should be near to a quartz, not far from the cpu and chipset)

[Gautam]

Franck, why is the CPU code name listed as "Nehalem" instead of "Bloomfield"?

Sorry to jump in with that question like that but it's been bugging me for a bit.

[cpuz]

Franck, why is the CPU code name listed as "Nehalem" instead of "Bloomfield"?

Sorry to jump in with that question like that but it's been bugging me for a bit.

yeah you're right, it should be Bloomfield.
Nehalem is the name of the micro-architecture.

[bowman]

Franck, why is the CPU code name listed as "Nehalem" instead of "Bloomfield"?

Sorry to jump in with that question like that but it's been bugging me for a bit.

There could be two reasons.. #1 ES chips are just called Nehalem, or #2 CPU-Z just doesn't recognize it properly yet. Notice how the 'FSB' is 'quad pumped' by default even though there are no such things.

[Hornet331]

I think this one way more accurate , furthermore the software is not ready yet ...


...

we dont know the Tjmax of nehalem, so each program is not worth the effort.

e.g. realtemp take 95°C (supposed penryn TJmax) vs 105°C TJmax core temp takes, thats why you see 36°C vs 46°C.

for penryn i belive the 95°C more then the 105°C, cause some guys measured it.

For nehalem its a compleat different story, but maybe you have a IR thermometer, so you can test it. Just look at the realtemp thread, of how to do it.

[savantu]

RightMark Memory Analyzer v3.80


...

You're running single channel with 1 stick of DDR3 1066 ?

If so , the results are impossible.If not , the IMC has low efficiency.

Option 3 : memory is running at 1600MHz in single channel mode.

Which one ?

[bowman]

Well, this is IMC, dual channel DDR2 800..

http://www.isarapix.org/pix70/1214415744.png

So, if that's single channel, uh. Either eye-popping or my setup is borked.

[Donnie27]

You're running single channel with 1 stick of DDR3 1066 ?

If so , the results are impossible.If not , the IMC has low efficiency.

Option 3 : memory is running at 1600MHz in single channel mode.

Which one ?

I'd bet on Single channel 1600MHz LOL! Wasn't all 3 channel capable of something like 26GB?

[Movieman]

I'd bet on Single channel 1600MHz LOL! Wasn't all 3 channel capable of something like 26GB?

All the SS I've seen of Nehalem have been in single channel mode so far.
Why I don't know but they are.
Maybe a bug or perhaps someone is keeping the systems handicapped until they want the real capabilities of this beast to be shown.
Yea, I heard in the neighborhood of 26GB also.
Cringes at the thought of writing the check for 6 sticks(or more) of DDR3-1333 ECC for a dual socket Gainstown..

[bowman]

Triple-channel 1333 will be capable of 32GB/s.

Ha, Movieman, didn't you see the dual socket motherboards with 18 DIMM slots?

[Movieman]

Triple-channel 1333 will be capable of 32GB/s.

Ha, Movieman, didn't you see the dual socket motherboards with 18 DIMM slots?

Yea, I saw them..A waste for my needs..1 gig a core more than covers anything I need..

[Donnie27]

Triple-channel 1333 will be capable of 32GB/s.

Ha, Movieman, didn't you see the dual socket motherboards with 18 DIMM slots?

Total of 32GB but usable 26GB just like 9 out 12 right now, right? Current DDR3 boards does 21GB and look how much is usable? Or am I misisng something?

[Blauhung]

yeah you're right, it should be Bloomfield.
Nehalem is the name of the micro-architecture.

well, Nehalem is the product name throughout the in fab manufacturing process. It only becomes Bloomfield or Gainstown in assembly.

[Mr.BSEL]

Is there any information on whether there will be improvement in games at max details.

[Gautam]

well, Nehalem is the product name throughout the in fab manufacturing process. It only becomes Bloomfield or Gainstown in assembly.

More specifically Nehalem refers to a single core, Bloomfield and Gainestown refer to the processor, consisting of four Nehalem cores in this case.

[virtualrain]

I'd bet on Single channel 1600MHz LOL! Wasn't all 3 channel capable of something like 26GB?

Single channel DDR3-1600 has peak theoretical bandwidth of 12.8GB/s which is exactly what Rightmark is reporting. DDR3-800 in dual channel also has the same bandwidth though.

[Anemone]

Ok maybe I'm confused, but thought Nehalem referred to the core, which in this case starts as a quad core, and Bloomfield refers to the whole platform, what people are referring to as the X58, only it's the X58+1366 Nehalem that make a "Bloomfield" platform?

Probably I'm wrong like usual, but that was how I understood the names.