Monday, October 13th 2008

Core i7 940 Review Shows SMT and Tri-Channel Memory Let-down

As the computer enthusiast community gears up for Nehalem November, with reports suggesting a series of product launches for both Intel's Core i7 processors and compatible motherboards, Industry observer PC Online.cn have already published an in-depth review of the Core i7 940 2.93 GHz processor. The processor is based on the Bloomfield core, and essentially the Nehalem architecture that has been making news for over an year now. PC Online went right to the heart of the matter, evaluating the 192-bit wide (tri-channel) memory interface, and the advantage of HyperThreading on four physical cores. In the tests, the 2.93 GHz Bloomfield chip was pitted against a Core 2 Extreme QX9770 operating at both its reference speed of 3.20 GHz, and underclocked to 2.93 GHz, so a clock to clock comparison could be brought about.

The evaluation found that the performance increments tri-channel offers over dual-channel memory, in real world applications and games, are just about insignificant. Super Pi Mod 1.4 shows only a fractional lead for tri-channel over dual-channel, and the trend continued with Everest Memory Benchmark. On the brighter side, the integrated memory controller does offer improvements over the previous generation setup, with the northbridge handling memory. Even in games such as Call of Duty 4 and Crysis, tri-channel memory did not shine.
As for the other architectural change, simultaneous multi-threading, that makes its comeback on the desktop scene with the Bloomfield processors offering as many as eight available logical processors for the operating system to talk to, it appears to be a mixed bag, in terms of performance. The architecture did provide massive boosts in WinRAR and Cinebench tests Across tests, enabling SMT brought in performance increments of roughly 10~20% with general benchmarks that included Cinebench, WinRAR, TMPGEnc, and Fritz Chess. With 3DMark Vantage, SMT provided a very significant boost to the scores, with about 25% increments. It didn't do the same, to current generation games such as Call of Duty 4, World in Conflict and Company of Heroes. What's more, the games didn't seem to benefit from Bloomfield in the first place. The QX9770 underclocked at 2.93 GHz, outperformed i7 940, both with and without SMT, in some games.

Source: PC Online
Add your own comment

91 Comments on Core i7 940 Review Shows SMT and Tri-Channel Memory Let-down

#1
Mussels
Moderprator
system memory isnt something i'd want left to something as unstable as analogue voltages. they'd be so prone to interference...
Posted on Reply
#2
niko084
Mussels said:

Video encoders are going to cream themselves, at least.
Eh, doubtful... Most people I know that are heavy into video encoding are running dual quads, and 8+ gb of ram. If you are really into it you can afford the system built for it, if you are doing it as something for yourself, the difference between a Quad and a Quad with HT isn't going to be enough I think, especially considering the over clock ability of the Core2 quads.
Posted on Reply
#3
DanTheBanjoman
Señor Moderator
niko084 said:
Eh, doubtful... Most people I know that are heavy into video encoding are running dual quads, and 8+ gb of ram. If you are really into it you can afford the system built for it, if you are doing it as something for yourself, the difference between a Quad and a Quad with HT isn't going to be enough I think, especially considering the over clock ability of the Core2 quads.
Though in a few months dual 6-core machines with htt will cost as much as the current dual quads. Of course, it'll be interesting to know how it scales. This should be where CSI should show its power.
Posted on Reply
#4
Mussels
Moderprator
niko084 said:
Eh, doubtful... Most people I know that are heavy into video encoding are running dual quads, and 8+ gb of ram. If you are really into it you can afford the system built for it, if you are doing it as something for yourself, the difference between a Quad and a Quad with HT isn't going to be enough I think, especially considering the over clock ability of the Core2 quads.
triple octas with HT and 16GB of ram will get to them soon enough!

HT is something that *will* apply to video encoding, as will the ram bandwidth. cant you see those graphs in the first post!



Yeah its not video encoding, but in most reviews i see the performance gains for video encoding, tend to be very similar to winrar/winzip encoding tasks.
Posted on Reply
#5
DarkMatter
Analog signals are totally contrary to the goal for what digital computing was created: efficiency. You want to use as few energy as you can and circuits are bound to some degree of inaccuracy. In fact digital in reality is not 0 or 1. It works as if anything above 0,75 is seen as one and anything below 0,25 is seen as 0. You can't make a whole chip run at an specified voltage, specially because every circuit has resistance, so while the first transistor gets 1v the last one can easily get only 0,8v (add to that the fact that different power supplies, mobos, etc. give different input voltages). You have to make your circuit with this in mind. Now imagine you add more states, how do you know which state is 0,36V THROUGHOUT your whole chip when the voltage for the same state can change almost by that amount? You can't and you would only have one option: increase your input voltage by an order of magnitude...
Posted on Reply
#6
niko084
DanTheBanjoman said:
Though in a few months dual 6-core machines with htt will cost as much as the current dual quads. Of course, it'll be interesting to know how it scales. This should be where CSI should show its power.
Indeed, but I think even with scaling it wont much matter comparing dual 6 core chips to dual quads... Unless someone manages 5ghz out of an air cooled xeon 5k series...:laugh:

I know when I moved from my e6750 @ 3.2 to my x3210 @ 3.2 my encoding speeds increased by about 50%..
Posted on Reply
#7
FordGT90Concept
"I go fast!1!11!1!"
Mussels said:
system memory isnt something i'd want left to something as unstable as analogue voltages. they'd be so prone to interference...
Both analog and digital are prone to interference. I think I said it wrong though. Analog and digital are two different means of sending the actual data on a wave. Digital is taking the wave and pulling specific information out of it (binary). Analog is converting the wave into a format that can be used, like what happens in a CRT.

Imagine the string I spoke of earlier having extra points of information (up to 16 so far) induced in to it. As long as you read the signal on the other end the same way it was produced, you'll end up with more information on the same stream of electrons in the same amount of time. So on a single sine, you can induce and extract any number of points of data from it.

How exactly they do this, I have no idea. I don't even know how deep the ODR and HDR go (controller to translator on stick or direct to memory modules). I do know it works as seen in the PS3.
Posted on Reply
#8
FordGT90Concept
"I go fast!1!11!1!"
DarkMatter said:
Analog signals are totally contrary to the goal for what digital computing was created: efficiency. You want to use as few energy as you can and circuits are bound to some degree of inaccuracy. In fact digital in reality is not 0 or 1. It works as if anything above 0,75 is seen as one and anything below 0,25 is seen as 0. You can't make a whole chip run at an specified voltage, specially because every circuit has resistance, so while the first transistor gets 1v the last one can easily get only 0,8v (add to that the fact that different power supplies, mobos, etc. give different input voltages). You have to make your circuit with this in mind. Now imagine you add more states, how do you know which state is 0,36V THROUGHOUT your whole chip when the voltage for the same state can change almost by that amount? You can't and you would only have one option: increase your input voltage by an order of magnitude...
This is about all I can find on the subject of how they do it:
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2341
The next technology that FlexIO enables is DRSL with LVDS (Low Voltage Differential Signaling), which is a technology similar to what Intel uses in the Pentium 4 to reduce power consumption of their high-speed ALUs. We will actually explain the technology in greater detail later on this week in unrelated coverage, but the basic idea is as follows: normally the lower the voltage you run your interfaces at, the more difficult it becomes to detect an electrical "high" from an electrical "low." The reason being that it is quite easy to tell a 5V signal from a 0V signal, but telling a 0.9V signal from a 0V signal becomes much more difficult. DRSL instead takes the difference between two voltage lines with a very low voltage difference and uses that difference for signaling. By using low signal voltages, you can ensure that even though you may have a high speed bus, power consumption is kept to a minimum. The technology isn't quite sophisticated enough to make the transition to the mobile world, but with some additional circuitry to dynamically enable/disable interface pins it would be quite easy to apply FlexIO to mobile applications of the Cell architecture.
Posted on Reply
#10
CDdude55
Crazy 4 TPU!!!
As ive said, as a gamer this shows even more why i am not getting Core i7.


Core i7 i think will be a really good thing among the video encoders and photoshop people(and folding), for gaming they arent that great, even if the apps arent multi threaded, that still isn't much of an excuse as to why it cant beat the QX9770, with the apps of today.

Maybe when games use more cores, core i7 will be useful but even then i dont see a huge leap over the Core 2's.

Hearing this is even more disappointing news for the gamers.
Posted on Reply
#11
niko084
I think the I7 will have the market that they seem to be pushing it to..

Professional workstations... They are not overclocked and they need to be powerful and stable.
Posted on Reply
#12
DarkMatter
That's XDR again, which has nothing to do with SDRAM memory and it's neither analog by any means. It's neither octal data rated in the same way as DDR SDRAM is dual data rated.

It just uses 8 bit wide lanes to achieve 8 bits per clock per lane, but I fail to see how that is octal pumped.

From the Cell (microprocessor) wiki article:
The system interface used in Cell, also a Rambus design, is known as FlexIO. The FlexIO interface is organized into 12 lanes, each lane being a unidirectional 8-bit wide point-to-point path. Five 8-bit wide point-to-point paths are inbound lanes to Cell, while the remaining seven are outbound. This provides a theoretical peak bandwidth of 62.4 GB/s (36.4 GB/s outbound, 26 GB/s inbound) at 2.6 GHz. The FlexIO interface can be clocked independently, typ. at 3.2 GHz. 4 inbound + 4 outbound lanes are supporting memory coherency.
It's a completely different technology, with it's pros and it's cons. It's been long since DDR SDRAM probed to be the better economical RAM solution while providing almost the same performance as Rambus memory. XDR is better for embedded solutions that don't require too much memory.

I also still fail to see how you could do quad pumped SDRAM, that is that each memory cell performs 4 ops per clock cycle. And I also don't understand what would be the benefit of that, versus a DDR RAM with double the speed. I.e if your memory cells can perform 1600MT/s wouldn't it be better (simpler, easy to implement, cheaper...) a DDR running at 800Mhz than a "QDR" at 400Mhz?
Posted on Reply
#13
BOSE
Would i want to see.

What i really want to see, is multiple programs running at once. Tri-Chinal might not be good for games or most other applications, but perhaps it will do well in a multitasking world, where people are running at least 4 programs at once.

We need multitasking benchmarks.
Posted on Reply
#14
masterbw2000
This review is flawed.

They did not use Tri-Channel certified kit to do the test. The system was running at Dual channel mode although three are used.

Tri-channel REQUIRE Tri-Channel Certified Memory Kit like the A-DATA news we have seen previously.
http://www.techpowerup.com/72926/A-Data_Releases_Tri-Channel_Memory_Kits_for_Intel_X58_Platform.html

You cannot just buy 2 x Dual Channel Kit and grab 3 out of 4 sticks because that will not enable true Tri-Channel mode.
Posted on Reply
#15
BOSE
masterbw2000 said:
This review is flawed.

They did not use Tri-Channel certified kit to do the test. The system was running at Dual channel mode although three are used.

Tri-channel REQUIRE Tri-Channel Certified Memory Kit like the A-DATA news we have seen previously.
http://www.techpowerup.com/72926/A-Data_Releases_Tri-Channel_Memory_Kits_for_Intel_X58_Platform.html

You cannot just buy 2 x Dual Channel Kit and grab 3 out of 4 sticks because that will not enable true Tri-Channel mode.
No you dont. Its all the same RAM. Its just a selling gimmick for people like you.

You can either buy 3 packs of of 2x1GB or, you can buy 2 packs of 3x1GB sticks. Its all the same.
Posted on Reply
#16
niko084
masterbw2000 said:
This review is flawed.

They did not use Tri-Channel certified kit to do the test. The system was running at Dual channel mode although three are used.

Tri-channel REQUIRE Tri-Channel Certified Memory Kit like the A-DATA news we have seen previously.
http://www.techpowerup.com/72926/A-Data_Releases_Tri-Channel_Memory_Kits_for_Intel_X58_Platform.html

You cannot just buy 2 x Dual Channel Kit and grab 3 out of 4 sticks because that will not enable true Tri-Channel mode.
Although that is "certified" I highly doubt that it is "required"...

First of all there is no way for the computer to tell if its certified or not, and I doubt its an entire different line of ram..

It's like dual channel certified kits, you don't need that either, but they said you did.
Posted on Reply
#17
masterbw2000
BOSE said:
No you dont. Its all the same RAM. Its just a selling gimmick for people like you.

You can either buy 3 packs of of 2x1GB or, you can buy 2 packs of 3x1GB sticks. Its all the same.
Hello. There is no need to argue about this because I had doubts myself about that too previously.
Or you can say that both MB & RAM Manufacturers need to Cross-Certify because MB BIOS is involved in that as well.
But of course, you are most welcome to check with ASUS, Gigabyte, MSI, or any manufacturer that is going to sell the X58 boards. :)
Posted on Reply
#18
Morgoth
so what point are you making?
Posted on Reply
#19
DarkMatter
You don't need any kind of certification is all marketing BS.
Posted on Reply
#20
niko084
masterbw2000 said:
Hello. There is no need to argue about this because I had doubts myself about that too previously.
Or you can say that both MB & RAM Manufacturers need to Cross-Certify because MB BIOS is involved in that as well.
But of course, you are most welcome to check with ASUS, Gigabyte, MSI, or any manufacturer that is going to sell the X58 boards. :)
That board manufacturers also list "certified" memory... Yet, hmm almost everything works..

It's multi level marketing and cross marketing, if Asus says Crucial ram, Micron pays them.
Posted on Reply
#21
Unregistered
masterbw2000 said:
Hello. There is no need to argue about this because I had doubts myself about that too previously.
Or you can say that both MB & RAM Manufacturers need to Cross-Certify because MB BIOS is involved in that as well.
But of course, you are most welcome to check with ASUS, Gigabyte, MSI, or any manufacturer that is going to sell the X58 boards. :)
wait a minute , who do you work for ?

since you are from Taipei, Taiwan, maybe you have some of these samples ? you may be right , but that is a big maybe

I also think its marketing BS
Posted on Edit | Reply
#22
BOSE
It just makes it easier for people to buy ram in a pack of 3, instead of buying a pack of 2 and then an additional stick of ram.

Its common sense. This way, an average Joe or Jane feels better about their purchase because it has word "certified" on the package.

So let it go.
Posted on Reply
#23
masterbw2000
I can tell you that X58 sample boards are out there enjoying being "pumped". The launch date is coming up soon.

Notes:
1. Three (3) sticks out of Dual Channel Memory kits, test didn't show any gain.
2. Certified version of Tri-Channel memory shows large performance gain.
3. Circuitry differences.
Posted on Reply
#24
Morgoth
these is not differend in a dual pack or a tripack there both pysical and technical the same
if i buy 3x 1gb single or 3 paks of 2x2/2x1 or a 3 pack it all runs trichannel can also run dual channel
i wil profe to you that 3x dual channel pack can run tri channel once i got my stuff
Posted on Reply
#25
niko084
Morgoth said:
these is not differend in a dual pack or a tripack there both pysical and technical the same
if i buy 3x 1gb single or 3 paks of 2x2/2x1 or a 3 pack it all runs trichannel can also run dual channel
i wil profe to you that 3x dual channel pack can run tri channel once i got my stuff
You can also verify that its running in dual or tri channel in the bios as well as the os.
Posted on Reply
Add your own comment