Yep! --- DDR3 is nothing really special, rather hitting the JEDEC spec of 1.5 V at a state rate... mem companies take a shortcut rather than actually hitting this spec, they just pump up voltage to the hairy limit.Originally Posted by XS Janus
Yep! --- DDR3 is nothing really special, rather hitting the JEDEC spec of 1.5 V at a state rate... mem companies take a shortcut rather than actually hitting this spec, they just pump up voltage to the hairy limit.
One hundred years from now It won't matter
What kind of car I drove What kind of house I lived in
How much money I had in the bank Nor what my cloths looked like.... But The world may be a little better Because, I was important In the life of a child.
-- from "Within My Power" by Forest Witcraft
due to the cache structure no?
" Business is Binary, your either a 1 or a 0, alive or dead." - Gary Winston ^^
Asus rampage III formula,i7 980xm, H70, Silverstone Ft02, Gigabyte Windforce 580 GTX SLI, Corsair AX1200, intel x-25m 160gb, 2 x OCZ vertex 2 180gb, hp zr30w, 12gb corsair vengeance
Rig 2
i7 980x ,h70, Antec Lanboy Air, Samsung md230x3 ,Saphhire 6970 Xfired, Antec ax1200w, x-25m 160gb, 2 x OCZ vertex 2 180gb,12gb Corsair Vengence MSI Big Bang Xpower
highest ddr3 dimm that i know with 1.5V is DDR3 1333 cl7. But its muy expensive (220€)....
Ummm, for an enthusiast line of processors that have their own mobo type, shouldn't ...ahem... Intel have tested this basic feature?
How do you know they havent?
fry your Nehalem with some performance ram:
http://www.slipperybrick.com/2008/10...2ghz-ddr3-ram/
The HyperX PC3-16000 memory is part number KHX16000D3K2/2GN and has timings of CL9-9-9-27 at 1.9 volts. Pricing for the modules is $227 for a kit of two 1GB sticks for a 2GB kit. Kingston also backs the RAM with a lifetime warranty and 24/7 tech support.
i7 3610QM 1.2-3.2GHz
I've always liked your avatar
One hundred years from now It won't matter
What kind of car I drove What kind of house I lived in
How much money I had in the bank Nor what my cloths looked like.... But The world may be a little better Because, I was important In the life of a child.
-- from "Within My Power" by Forest Witcraft
Ha ha jack, seems like you are spending more time here than on xcpus.I've always liked your avatar
Anyhow, is this going to just be a temporary thing untill they get a new revision out, or forever more?
Core i7 920 D0 B-batch (4.1) (Kinda Stable?) | DFI X58 T3eH8 (Fed up with its' issues, may get a new board soon) | Patriot 1600 (9-9-9-24) (for now) | XFX HD 4890 (971/1065) (for now) |
80GB X25-m G2 | WD 640GB | PCP&C 750 | Dell 2408 LCD | NEC 1970GX LCD | Win7 Pro | CoolerMaster ATCS 840 {Modded to reverse-ATX, WC'ing internal}
CPU Loop: MCP655 > HK 3.0 LT > ST 320 (3x Scythe G's) > ST Res >Pump
GPU Loop: MCP655 > MCW-60 > PA160 (1x YL D12SH) > ST Res > BIP 220 (2x YL D12SH) >Pump
I go back and forth ...
Regarding your question I have no idea ... I posted a link showing TDDB and lifetime measurements for high-k related research in another thread, my suspicion (as it appears many players, including IBM and AMD) is that the industry will ultimately implement this material in many different applications that drive DRAM and the industry will ultimately drive the memory makers to actually improve their products. By this I mean, DDR3 is nothing especially revolutionary.... a small evolution over DDR2, and the high clock/low latency performance DRAM is, as usual, driven by nothing more than simply put voltage on it and hope it works ... those modules that do work, they slap a label on it and sell it. As soon as the JEDEC sets the standard, everyone launches something that brands way outside that standard.
Ultimately, the need for high performance memory at lower voltages will create the necessity for DRAM manufacturers to get more creative. A DRAM maker that finds a way to create a low cost memory technology at low voltage, high performance will become very successful in the long run.
Fred Weber (former AMD CTO) is just a brilliant technologist -- he was interviewed not too long ago, and when he left AMD he started a new memory company ... this makes me happy
Edit: Here's the link to the interview: http://www.theregister.co.uk/2008/02...d_weber_scc13/
Here's a link to his new company:
http://www.metaram.com/investors.html see if you can spy the logo
Last edited by JumpingJack; 10-03-2008 at 11:03 PM.
One hundred years from now It won't matter
What kind of car I drove What kind of house I lived in
How much money I had in the bank Nor what my cloths looked like.... But The world may be a little better Because, I was important In the life of a child.
-- from "Within My Power" by Forest Witcraft
Why is it that SuperPI has become such a popular benchmark for testing Intel CPU's in the last few years?
A yorkie at 3.33 Ghz does SP1M at 14.125 seconds.
Jack
SuperPi was popular since...ever.
It's not only for Intel, and we were also running it all the time with the Athlon XPs, Intel P4's and the Athlon 64s and some people with Dothans.
Coding 24/7... Limited forums/PMs time.
-Justice isn't blind, Justice is ashamed.
Many thanks to: Sue Wu, Yiwen Lin, Steven Kuo, Crystal Chen, Vivian Lien, Joe Chan, Sascha Krohn, Joe James, Dan Snyder, Amy Deng, Jack Peterson, Hank Peng, Mafalda Cogliani, Olivia Lee, Marta Piccoli, Mike Clements, Alex Ruedinger, Oliver Baltuch, Korinna Dieck, Steffen Eisentein, Francois Piednoel, Tanja Markovic, Cyril Pelupessy (R.I.P.), Juan J. Guerrero
There are lots of chips that can do 1800MHz under 1.7Vdimm. And yes, Nehalem IMC will bring better memory speed at same voltage, so nothing to worry about. Also by setting higher cpu voltage you can set higher Vdimm too. I don't see why is anyone shouting about this. Many people killed their CPU's back in K8 days with BH5.
Its actually pretty funny. Since Nehalem could destroy Core 2 in any terms of memory with its left leg broken.
Core 2 uses a unidirectional 64bit FSB to the MCH. So even at 2000FSB and whatever uber memory 1600Mhz DDR2. A trichannel Nehalem with DDR3-800 will have atleast 20% more bandwidth. And thats asuming you dont use the FSB for anything else.
Next is the latency. Since the MCH step is gone and a huge latency part with it. Then a CL9 chip would outperform a CL7 and so on compared with the Core 2. In short, you can now be just as badass on Nehalem with value memory that you could on Core 2 with the best possible memory out there.
Crunching for Comrades and the Common good of the People.
I'm not entirely understanding why the latency is lower in the end with Nehalem.
Is the route from memory to final destination just significantly shorter? I mean, it sounds like that's effectively what you're saying, but I have to question if there's not more to it. Just curious.
Last edited by Sly Fox; 10-04-2008 at 03:47 AM.
I don't think that there were MANY users that suffered a Athlon 64 / Opteron death from high Vdimm to be honest.
I myself had about 10 Athlon 64s ( various cores, Winchester, Venice, San Diego, Venus ) and 2 Opteron 146 and I've been using BH5 at 3.2V minimum, and usually at 3.7 to 4.0V and didn't lose a single CPU.
I was using those high RAM voltages with the default CPU's voltage as well as overclocked.
And we should also consider the possibility of a random death or something else degrading/dieing and that it may not be the high RAM voltage to blame.
I for one, like I already said ( posted to be precise) bet that I will be able to run many Nehalem systems with the CPUs @ stock clocks & voltages with the RAM at 2.27V without losing a CPU or degrading it.
I will do a similar to the current die-hard test ( C2Ds & C2Qs 45nm with high Vcore, or Vpll, or Vtt, or a combination of all three ) with the Nehalem soon.
I think I'll dedicate three systems to that purpose, running the RAM at 2.27V and F@H 24/7 for a month or even more to see what's going to happen.
Coding 24/7... Limited forums/PMs time.
-Justice isn't blind, Justice is ashamed.
Many thanks to: Sue Wu, Yiwen Lin, Steven Kuo, Crystal Chen, Vivian Lien, Joe Chan, Sascha Krohn, Joe James, Dan Snyder, Amy Deng, Jack Peterson, Hank Peng, Mafalda Cogliani, Olivia Lee, Marta Piccoli, Mike Clements, Alex Ruedinger, Oliver Baltuch, Korinna Dieck, Steffen Eisentein, Francois Piednoel, Tanja Markovic, Cyril Pelupessy (R.I.P.
There is a longer route by tracelengths alone. And adittionally the MCH is a buffered place with store and forward procedures.I'm not entirely understanding why the latency is lower in the end with Nehalem.
Is the route from memory to final destination just significantly shorter? I mean, it sounds like that's effectively what you're saying, but I have to question if there's not more to it. Just curious.
Crunching for Comrades and the Common good of the People.
Feel free to correct me if I am wrong, but I get the impression that SuperPi's importance or perhaps popularity is a better word, as a benchmark, has increased significantly since Conroe came onto the scene.
Is SuperPi one of the better predictors of across the board performance?
Across the board ? Nope. Not all.
Actually there's no single benchmark that can indicate a CPU's performance across the board.
SuperPi is just a simple number crunching benchmark that we ( the overclockers ) love to run and benchmark our systems with.
The only way to have an idea about the overall performance of a CPU is to put it through various benchmarks.
Edit: SuperPi became more popular ( among the average joe users ) with the arrival of the Core 2 CPUs because it's much easier now to crack down some nice scores than in the past.
To reach 24s in the past was a PITA, but now nearly everyone can nail down 12s with an air cooled Core 2 CPU.
Last edited by BenchZowner; 10-04-2008 at 05:41 AM.
Coding 24/7... Limited forums/PMs time.
-Justice isn't blind, Justice is ashamed.
Many thanks to: Sue Wu, Yiwen Lin, Steven Kuo, Crystal Chen, Vivian Lien, Joe Chan, Sascha Krohn, Joe James, Dan Snyder, Amy Deng, Jack Peterson, Hank Peng, Mafalda Cogliani, Olivia Lee, Marta Piccoli, Mike Clements, Alex Ruedinger, Oliver Baltuch, Korinna Dieck, Steffen Eisentein, Francois Piednoel, Tanja Markovic, Cyril Pelupessy (R.I.P.
One question, has somebody really destroyed Nehalem with high Vdimm?
Though I understand the rhetorical meaning of your question ... and I don't disagree. That is why the benchmark should be put into context. It is not a good benchmark to compare against other architectures as a general performance benchmark, thus because AMD does not do well at SP vs Intel does not lend to making a general statement about the two products.
What SuperPI does do is exercise just a few of the many facets of a CPU. What good is it then? Well, it does a few things -- it helps ascertain the scaling of a CPU and drives straight to isolate only key components. It is not BW sensitive, it is not cache sensitive (this 'it runs all in cache crud is a bunch of bunk when using it to analyze a CPU').
SuperPI is inherently recursive, thus a huge loop with a conditional clause of 'when to stop'. Intel does well in this benchmark because they have strong logic in branch predictors and loop detectors. Intel has mentioned specifically that they reworked the loop detection logic in Nehalem, so seeing this improvement clock for clock in Nehalem is a good indication that some of the core tweaks made an impact.
The second thing SuperPI is good at, because it is so simple and straight forward, is to validate CPUID and clock speed. SuperPI should scale well, regardless of the architecture tested. As such, with a good baseline of data, a CPU clock can be extrapolated from the SuperPI run -- if they don't agree (superPI and CPUID) then it is reasonable to suspect something is 'fishy'.
One hundred years from now It won't matter
What kind of car I drove What kind of house I lived in
How much money I had in the bank Nor what my cloths looked like.... But The world may be a little better Because, I was important In the life of a child.
-- from "Within My Power" by Forest Witcraft
I am sure it has been done .... people have reported destroying a Phenom with high Vdimm as well, though the ceiling is much much higher.
I have pretty much convinced myself from what i have been able to research in the literature and such that the voltage sensitivity in Nehalem is related to the high-K process Intel uses.
Last edited by JumpingJack; 10-04-2008 at 07:17 AM.
One hundred years from now It won't matter
What kind of car I drove What kind of house I lived in
How much money I had in the bank Nor what my cloths looked like.... But The world may be a little better Because, I was important In the life of a child.
-- from "Within My Power" by Forest Witcraft
I don't take this man's word lightly, so the answer must be yes:
Yeah, basically you are on the right track. In the current off-die architecture the data from memory to CPU is two hops away from the CPU, in an IMC configuration it is only one hop away.I'm not entirely understanding why the latency is lower in the end with Nehalem.
Is the route from memory to final destination just significantly shorter? I mean, it sounds like that's effectively what you're saying, but I have to question if there's not more to it. Just curious.
You can study up on the affect of 'electrical distance' overall just by looking up the effective latency of an AMD dual socket platform today. In fact, this is why they call AMD's current implementation (and now Intel's future implementation) NUMA, since in a 2 socket server, any one CPU may have 1/2 it's memory 2 hops away.... i.e. memory access is non-uniform, hence the name non-uniform memory access (NUMA).
The concept is that a CPU has both a local (memory directly attached) and remote access (memory located on another CPU accessed via the HT link and soon with intel the QPI link). The impact is that when the CPU needs data from memory located on another CPU there will be extra latency associated with the access/call to that memory --
ftp://service.boulder.ibm.com/eserve...s21_081506.pdf (section 2.2 figure 3).
Jack
One hundred years from now It won't matter
What kind of car I drove What kind of house I lived in
How much money I had in the bank Nor what my cloths looked like.... But The world may be a little better Because, I was important In the life of a child.
-- from "Within My Power" by Forest Witcraft
QPI is essentially HT, its all based off the same technology and as such suffers from the same disadvantages when it comes to high vdimm. Phenom when first released had some serious limits to Vdimm, i7 has the same limits and for the same reasons...the thing you guys keep missing though is you may NOT need the ubber high vdimm and I for one will not be fricken interested in running 2000MHZ on it either, for me the thought of low latencies tri channel is way to tempting, 1333 6-6-6- or even 5-5-5- if the MCH supports cas5 is way to mouthwateringly good to pass up. Also we are done with TRD (thanks the Intel gods) and the tricks motherboard manufacturers played to get boards doing silly FSB...as if FSB at 2400 matters :shakeshead:
I think we are dawning on a new fresh memory overclocking day with i7...yes some will push 2000mhz, some will think the read bandwidth is the be all and end off of benches, those who are clever wil look to real tight access latency and huge write/copy speed.
1333 5-5-5- or 6-6-6- is where I hope to be, or even 6-5-5-
I have the boards, just need the CPU now
Got a problem with your OCZ product....?
Have a look over here
Tony AKA BigToe
Tuning PC's for speed...Run whats fast, not what you think is fast
Well said.QPI is essentially HT, its all based off the same technology and as such suffers from the same disadvantages when it comes to high vdimm. Phenom when first released had some serious limits to Vdimm, i7 has the same limits and for the same reasons...the thing you guys keep missing though is you may NOT need the ubber high vdimm and I for one will not be fricken interested in running 2000MHZ on it either, for me the thought of low latencies tri channel is way to tempting, 1333 6-6-6- or even 5-5-5- if the MCH supports cas5 is way to mouthwateringly good to pass up. Also we are done with TRD (thanks the Intel gods
I think we are dawning on a new fresh memory overclocking day with i7...yes some will push 2000mhz, some will think the read bandwidth is the be all and end off of benches, those who are clever wil look to real tight access latency and huge write/copy speed.
1333 5-5-5- or 6-6-6- is where I hope to be, or even 6-5-5-
I have the boards, just need the CPU now
One hundred years from now It won't matter
What kind of car I drove What kind of house I lived in
How much money I had in the bank Nor what my cloths looked like.... But The world may be a little better Because, I was important In the life of a child.
-- from "Within My Power" by Forest Witcraft
Posting Permissions
Reply With Quote
Bookmarks