Information

Did you know you could become an X-bit labs fan on Facebook or follow us on Twitter?



Discussion

Discussion on Article:
Intel Core i7-3820 vs. Core i7-2700K and Core i7-3930K

Started by: BestJinjo | Date 01/29/12 08:58:26 AM
Comments: 27 | Last Comment:  10/22/12 10:46:23 AM

Expand all threads | Collapse all threads

[1-7]

1. 
Excellent review.

Once Ivy Bridge launches, even LGA1155 platform will have more than sufficient bandwidth for videocards courtesy of PCIe 3.0 in Ivy Bridge CPU.

Secondly, Ivy Bridge should bring anywhere from 7-18% more performance per clock in applications as a result of architectural improvements and a much more aggressive Turbo Boost 2.0:

http://www.cpu-world.com/..._Bridge_desktop_CPUs.html

and

http://www.nordichardware...dge-3770k-benchmarks.html

Thirdly, it will have even lower power consumption.

Fourthly, it will likely overclock higher than 4.9ghz you achieved on your 2700k, completely negating the purpose for an i7-3820.

In other words, in just a matter of months, Ivy Bridge will basically obsolete the entire LGA2011 platform, except for those who actually need 6 cores / 12 threaded CPU for web development, video encoding, encryption and rendering.

To me, the only reason an extremely expensive platform is justifiable (> $500 per CPU + $250 for mobo) is if that platform performs much faster in all situations, not just in 3-4 specific situations. A computer user can simply buy a $225-332 Ivy Bridge quad-core CPU and reallocate the savings from LGA2011 platform towards a faster GPU (or say a 2nd GPU) or a larger SSD, almost always making such a decision better than going with the overpriced LGA2011 platform.

Worst of all, Intel's "high-end" platform keeps launching after their mainstream platform. So it will take another 1-2 quarters before Intel releases Ivy Bridge-E. Intel should have just called LGA2011 a workstation/server platform and never called it their high-end platform. With the launch of Ivy Bridge, there will be nothing high-end about a $550 3930 CPU that will get whooped by a 5.2-5.3ghz quad core Ivy Bridge in almost all situations that apply to common users (office tasks, games, Photoshop, etc.).

And then the cycle will repeat again. Intel will launch IVB-E in Q3 2012, but then one might as well wait for LGA1150 and much faster Haswell. In my opinion, the strategy behind LGA2011 is completely inconsistent. At least LGA1366 launched a full year earlier, allowing early "high-end" adopters to claim exclusive performance over mainstream users.
2 0 [Posted by: BestJinjo  | Date: 01/29/12 08:58:26 AM]
Reply
2. 
Article says that the cost of an3820 system is $100 more than a 2700k system because X79 mobos are, on average, $100 more expensive.
I guess the author forgot 2700k is about $90 more expensive than 3820...
That brings the platform cost to an even field, and I think is something that affects the value of that platform considerably.
1 1 [Posted by: Tomash  | Date: 01/29/12 09:30:30 AM]
Reply
- collapse thread

 
Not really since 2600k overclocks exactly the same as 2700k and you are back to square 1: 2600k is still better for price/performance than 3820. 2700k actually makes no sense.

Most importantly, people have been using 2600k since January of 2011. Implying that 3820 is suddenly relevant misses the fact that it's 12 months late. But in that case, in just 3 more months, 3820 will be completely obsoleted by 3770k. It's actually impossible to make any argument for 3820 today, not when IVB is going to launch by 2nd week of April.
2 1 [Posted by: BestJinjo  | Date: 01/29/12 06:15:52 PM]
Reply
 
A 3820 is a sandy E chip that was supposed to come out after the Sandy Bridge chips. How is it 12 months late?
0 0 [Posted by: veli05  | Date: 02/16/12 06:43:13 AM]
Reply
3. 
All the comparisons between the 2011 and 1155 platforms I've seen neglect one important thing.

"longevity".....

The last shout of the 1155 platform will be a quad core IvyB with hyperthreading and 8MB shared L3 cache.

Whereas the last shout on the 2011 platform will be an octa core IvyB-E with hyperthreading and 20MB of shared L3 cache.

There's no doubt when you consider the above that the 2011 platform will remain viable for substantially longer than the 1155 platform, particularly when you consider the increasingly multi threaded software environment.

My plan is to initially go with a 3820 on lga2011 and upgrade to an octa core IvyB-E on the same platform when prices drop on the release of Haswell-E by which time octa core will be more beneficial as software becomes progressively more multi threaded.

Now that's what I call longevity.
3 1 [Posted by: technogiant  | Date: 01/29/12 10:57:24 AM]
Reply
- collapse thread

 
That is assuming, that current mobos will be compatible with Haswell. This is far from certain, and given Intel's track record, when it comes to backwards compatibility, I'd lean to the "not happening" side.
2 1 [Posted by: rrr  | Date: 01/29/12 01:12:19 PM]
Reply
 
No..not talking about compatibility with Haswell....it won't be. You misunderstood what I wrote....I'm assuming compatibility of IvyB-E on the current 2011 platform...which from leaks will probably be the case...although obviously not confirmed by Intel at this time.
2 1 [Posted by: technogiant  | Date: 01/29/12 01:28:32 PM]
Reply
 
That's awfully trusting of the same Intel that still keeps 6-core processors for socket 1366 at the $600 mark, even as 1366 is considered a dead platform.

I doubt the 8-core HT enabled Ivy chips will ever be affordable, or at least a compelling price/performance since Intel will just roll Haswell out and again euthanize their prior platform if they inconveniently compete with new chipset sales.

Pretty sure Sandy Bridge could have been easily adapted to run on socket 1366 too, but again... gotta keep those margins. Plus there's not enough pressure from AMD in the high end to really threaten Intel there.

And ugh, I actually feel sorry for socket 1156 owners - that was just brutal.

Just my 2 cents.
2 2 [Posted by: xrror  | Date: 01/29/12 03:05:28 PM]
Reply
 
Longevity? Enthusiasts don't particularly care to keep their platforms for 4-5 years. You can have your 8-core IVB-E (but that hasn't even been confirmed). Even if true, most enthusiasts who buy 2500k/2600k and overclock them to 4.7ghz+ will upgrade to Haswell in 2013. Haswell will bring another 15% or so IPC improvement, making IVB less high-end once more. So the longevity factor of LGA2011 is a non-factor. The only benefit of LGA2011 platform is for people who actually need to use 6 C/ 12T or more. For example, an automotive company that does airflow simulation for the engine, etc. Those people don't browse Xbitlabs, they just go out and buy a $5,000 system every 2-3 years based on their company's budget.

I think Xbitlabs is coming from an enthusiast's point of view. For us enthusiasts, the LGA2011 platform is a huge disappointment since the CPUs cost 2-3x more, have worse power consumption and don't even come with much higher stock clocks or overclock better. Therefore, their only use is if you run programs that need more than 4 full cores.

At least LGA1366 came out earlier, giving enthusiasts real bragging rights. i7 920/930 also overclocked a little bit better than i7 860 (and weren't gimped by lower Vtt limitations of Lynnfield).
1 1 [Posted by: BestJinjo  | Date: 01/29/12 06:16:51 PM]
Reply
 
You make a good point there, and of course Haswell will have an improved architecture and higher clocks, but the higher core count on IvyB-E will help to negate that, more so as software parallelism increases.
I remember going through a similar thought process when trying to decide on my current build, whether to go for the quad core Q6600 or faster clocking Wolfdale dual core. In the end I decided on the Q6600 and 4-5 years later its still going strong for most games (@3.6GHz)and I'm pleased I did as I doubt the dual core would have had the same longevity. I'm just appplying the same logic here with 2011 vs 1155.
There is after all little point in being right on the bleeding edge in terms of cpu for gaming which is my primary concern when most games are gpu bound at resolutions actually used in real life. It's far better to lose a little in terms of IPC and clock speed in favour of extra cores which will ultimately provide more benefit as games leverage them. Many games are now starting to use quads fully and the 6 - 8 cores of SB-E and IvyB-E will be of more use than a few percent extra on IPC and clocks.
Perhaps you would not class me as an enthusiast, but I simply don't have the budget to replace my whole platform annually or biannually and would prefer to pay more initially for something that will last 4-5 years.
1 0 [Posted by: technogiant  | Date: 01/30/12 12:27:27 AM]
Reply
 
The reason Q6600 lasted you so long is not only because it's a quad-core. It's because most games have been GPU-limited for the longest time. Q6600 might have helped you in BF2 or Resident Evil 5 or GTAIV that do benefit from a quad-core processor. But what happens in a game that needs fast cache and IPC?

Q6600 flops hard:
http://www.techspot.com/r...2-performance/page13.html

In other words, by the time we'll need to use more than 4 full cores in a game, 6-core IVB will be very slow since we'll be on processors past Haswell by that point (think 2015+).

In addition to that, you'll be missing out on all the other new features that are yet to come such as the next generation PCI Express after 3.0 that we'll need to get past the 600mb/sec limitation for SSDs, Thunderbolt, etc.

Future processors will overclock even better and combined with higher IPC, provide far faster performance.

The argument for future proofing almost never makes sense since enthusiasts resell their hardware. Why spend $600-1000 + $300 on an LGA2011 platform to "future-proof"? You are basically paying a lot of $ upfront today for almost no benefit in hopes that down the line in 3-4 years programs start to use 6-8 threads fully. But by then we'll have even faster processors.

The same thing happened to Q6600. I owned one. When programs started to benefit from 4 cores, I switched to the much faster Sandy Bridge anyway because Q6600 became a huge bottleneck in games that needed fast IPC (such as Starcraft 2). We aren't even anywhere near the point where most programs use 4 cores. That means in the next 2-3 years, PCs will enter the "golden age" when quad-cores will finally be fully utilized. This means that a very high clocked, high IPC Quad-Core processor is actually going to be the sweet spot. The rest of the $ is better spent on an SSD/GPU/larger monitor, etc.

Think about this.

Q6600 (Kentsfield) to Q9550 (Wolfdale) is 5% increase in IPC.
Q9550 to i7 920/860 (Nehalem / Lynnfield) is 15% increase in IPC.
i7 920/860 to 2500/2600K (Sandy) is 15% increase in IPC.
Ivy will probably bring at least a 5% increase in IPC over Sandy Bridge.

22nm Ivy will probably overclock to at least 5.0ghz.

So a 5.0ghz IVB * 1.05 * 1.15 * 1.15 * 1.05 = 7.29ghz Kentsfield (Q6600).

In other words 2x faster than your CPU. The exact same thing will happen to IVB as future processor launch.
1 0 [Posted by: BestJinjo  | Date: 01/30/12 08:34:42 AM]
Reply
 
Lol...its interesting to see different view points. I think you sum it where you said
The reason Q6600 lasted you so long......It's because most games have been GPU-limited for the longest time
..This is true and there is no reason to believe that this is going to change any time soon.
I won't argue that the Q6600 has not had its day but it's note worthy that you used "starcraft 2", a game renowned for being cpu limited and an anomaly among games a to prove the point.
Also the fact that even though Q6600 is now 1/2 the performance per core of some modern day cpu's and yet is still able to run the vast majority of games more than adequately really only goes to prove my point that having the latest in bleeding edge ipc and cpu clocks is not essential and "future proofing" by ensuring you have sufficient cores to take advantage of more threaded future titles is also to be considered.
Really the only reason I'm considering upgrading is because the additional cpu load of running a multi gpu setup is taxing my cpu too much and bottle necking their performance.
Also I'd like to play Skyrim but understsand that more due to poor coding than anything else this requires a higher performing cpu.

So to cut to the chase I certainly think that sacrificing higher clocks was the correct choice in terms of choosing the Q6600 over a dual core.
I accept that whether this would be the correct view to take going forward is debatable, particularly when you consider that "going wider" is an area of diminishing returns while increased ipc's and clocks are not....although I don't think quad cores represent that "going wider limit" and think that in the life span of my new build there will be several games taking advantage of >4 cores.
0 0 [Posted by: technogiant  | Date: 01/30/12 10:42:12 AM]
Reply
 
I think Q6600 was one of the best CPUs though, esp. since it didn't have a very large price premium over Wolfdales. At the time I also went with it instead of E8400-E8600 Wolfdales. I got the Q6600 for $300. At the time E8500 was $299 and E8400 was $249:
http://www.anandtech.com/show/2468/2

Now look at the difference between 2500k/2600k and the cheapest 6-core: $300-400 (because 2500k can be purchased for $225 and the cheapest 3930 usually hovers around ~ $600). The delta grows even more if you add the $100+ motherboard premium. That's insane esp. since in 2 months IVB will make all LGA2011 CPUs slower in 99% of games that still struggle to fully use 4 cores.

The reason I used SC2 is because it's very popular and one of the most CPU limited games. It also shows us what happens when the game uses less than 4 cores. It is a game that shows how important IPC and frequency is. In games where CPU speed doesn't help, you are no better off with a $999 3960 than you are with an i3 2100 which is a $125 CPU.

But there are plenty of other games like SKYRIM:

http://www.techspot.com/r...im-performance/page7.html

and

http://www.tomshardware.c...apu-benchmark,3120-3.html

I think you are making a huge mistake if you are buying LGA2011 to futureproof for games. I can't see any game in the next 2-3 years where a fully clocked 3930 will be faster than a fully overclocked 3570K/3770K.

It looks like the next Xbox 720/Loop and PS4 won't launch until Fall 2013 at the earliest. It generally takes developers another 2-3 years to start making next generation game engines to take advantage of new consoles. By that time a game which uses 6 threads well (i.e., benefits from 3930) that CPU will be too slow vs. modern CPUs of 2015-2016. Basically, you'd probably be better off getting a 3570K ($225) or 3770K ($332) and in 2-3 years buying a brand new CPU platform than blowing $600 on 3930 + $300 on an LGA2011 today. You could spend the $400+ savings on a 2nd GPU, etc.

Look at i3 2100 vs. X4 955 or i5-2400 vs. FX-8120. Faster cores > more cores. That's unlikely to change esp. since we are only now entering a period where games start to use more than 2-3 threads. It's unlikely that suddenly games will start using 5-6 threads when we aren't even in the period where 4 threads are fully utilized.

But even worse, a $332 3770K will be faster than a $600 3930 in all games today. That's unacceptable given that it's almost 2x more expensive for the CPU alone. Also, an overclocked 3930 consumes way more power, requiring the use of a $60-80 high-end air cooler while 3570K/3770K will probably overclock past 5.0ghz on a $30 CM 212+ Evo.

http://www.nordichardware...dge-3770k-benchmarks.html

As to your comment: "...think that in the life span of my new build there will be several games taking advantage of >4 cores."

If you buy and hold your systems for 4-5 years, that in itself is a bad strategy. You are better off upgrading more frequently. I am actually amazed that someone as experienced as you still believes that buying and holding is better than upgrading every 2-3 years.
1 0 [Posted by: BestJinjo  | Date: 01/30/12 11:03:18 AM]
Reply
 
That's insane esp. since in 2 months IVB will make all LGA2011 CPUs slower in 99% of games


This is simply not the case, as most games are gpu limited not cpu limited....99% of games will perform identically on IvyB compared to SB-E and a very small number will perform slightly better on IvyB.

Look at i3 2100 vs. X4 955 or i5-2400 vs. FX-8120. Faster cores > more cores.


Again not the case, don't get drawn in by the cpu's being benchmarked with games at low resolutions and quality settings.....at usual game settings with normal gpu intensive games you will see little difference in fps on a faster cpu...to a point of course.

And as regards cost, the SB-E 3820 is meant to be $285 which in my country equals about £220 where as the 2500k=£160, 2600K=£250 and 2700K=£270 so I'd imagine IvyB being in the £300 region. So cost isn't loaded entirely on the side of 1155.

It's an interesting discussion and I can see merit on both sides.....its going to be an interesting couple of months ;-)
0 0 [Posted by: technogiant  | Date: 01/30/12 12:54:48 PM]
Reply
 
The benchmarks for SC2 and SKYRIM are at 1920x1080 with AA on. Don't forget to consider minimum framerates too.

Your own argument is inconsistent then. If you argue that for 99% of games, CPU speed doesn't matter, than why would you want to overspend for a 6-core CPU then? In that case, an i5 2400 is just as good as a $999 3960. Then you state that there are very few games where CPU speed actually matters. But it is precisely in those games where IPC and high overclocked frequency wins over more slower cores.

At the end of the day, it's your $. Feel free to buy a 6-core IVB-E on LGA2011 to futureproof. Not sure why you assumed IVB will be 300 pounds though as it will replace 2700k at the same price point. IVB will be superior in every way imaginable to the 3820. Higher overclocking, higher IPC and low power consumption. Actually, if your argument is that more cores is better for futureproofing, then the only CPU that makes sense for LGA2011 is the 3930 since it gives 6 cores and overclocks as good as the 3960. It is the 3820 that makes the least sense since then you can't even use "more cores is better viewpoint".

I also think in the next 1-2 years as GPUs get much more advanced, but developers are still focused on Xbox360/PS3 as the lowest denominator, most games will become much more CPU limited where a fast quad-core CPU will be more appealing than a slower 6-core CPU.

Also, in 2012, 2 very cpu limited games are coming out: (1) SC2 Heart of the Swarm and (2) Diablo 3. So even more reason to want a fast IPC/high overclocking quad.

Next year Haswell is going to bring support for new instructions. It's pretty difficult to make the future-proof argument when already next year Intel is going to have a brand new architecture:

http://www.bit-tech.net/n...tels-first-haswell-chip/1

IVB-E is nowhere near on the horizon for LGA2011.
0 0 [Posted by: BestJinjo  | Date: 01/30/12 02:59:04 PM]
Reply
 
To start both SC2 and Skyrim are bad examples, they are cpu limited even at 1920x1080 and as said before not the norm.

As regards my argument being inconsistent, I don't think so. For gaming, cpu performance provided it is within an acceptable range is not unimportant but is not as important as gpu. None the less when making a new system I would not go for an i5 2400 as it would not remain in that acceptable range of performance for long enough. To that end I'd go for the most powerful cpu I could afford within budget as it would give the system greater longevity.

With that in mind the i7 3820 is by far the best purchase if the $285 price tag is to be believed, and currently a quad is all that is required.
Even if the X79 platform is more expensive that is negated by the fact that I'd be able to upgrade to a 6-8 core IvyB-E in a couple of years when I still maintain those extra cores will be of use and the IvyB-E relative performance would still be well within that acceptable range.

Another advantage of IvyB-E that we haven't mentioned much is the huge shared L3 cache...20MB...even in games that don't use all its cores, those cores that are running will have access to all the L3 which would give it considerable advantage over standard IvyB with shared 8mb L3.
0 0 [Posted by: technogiant  | Date: 01/30/12 10:15:09 PM]
Reply
 
I guess you are still not understanding what I am saying. SC2 and SKYRIM are perfect for testing CPU limitation because we want to know in which games CPU speed actually matters. Ignoring them misses the entire point being discussed.

If we just tested Crysis or Witcher 2, then you'd be almost entirely GPU limited, in which case it there is no argument to be made for any CPU faster than i5 2400.

Also, you should come back in 2-3 years and see what happens. I guarantee you that Haswell (and possibly even its refresh) will demolish your IVB-E in 2 years in 99% of games. By the time you decide to upgrade your 3820 to IVB-E, the extra cores won't help you at all, just like Q6600's cores don't help it against modern CPUs because each of the cores is much slower.

By that time, we'll be on 14nm refresh of Haswell with ridiculously high clock speeds and even higher IPC.

Either way, in just 2 months time, come back to this site when 3770K beats 3820 in all the gaming benchmarks, and that's before overclocking is considered.

Even looking at your Q6600, it shows your theory/strategy doesn't work at all. Even a stock dual core i3 2100 beats Q6600 and Q6600 @ 3.4ghz has no hope at all to catch up to a stock i5 2400 in any CPU limited game out today.

Unless games start to use 6-8 threads well soon, 6 cores are going to be useless for videogames for a long time. Based on how long it's taking games to start taking full advantage of a quad-core CPU (Q6600 came out in Q1, 2007), it will take another 3-4 years before games use 6+ cores. By that point, a $600 IVB-E will be ancient news. Regardless, 3770K has HT. That's enough "future-proofing" as it is.

If we look at what happened in the last 10+ years, your CPU future proofing strategy has never worked. For instance, when Athlon X2 3800+ had 2 cores and Athlon 4000+ was a single core, by the time games started to use 2 full cores, we were already onto Core 2 Duos with 3.2-3.4ghz overclocks, making even the "better" X2 3800+ obsolete. And in just 2 months, Q6600 @ 3.4-3.6ghz will be 2x slower than a quad-core IVB @ 5.0ghz. Looking at FX-8120 and 8150 CPUs, it's already evident that more cores is not sufficient to overcome their inefficiency.
0 0 [Posted by: BestJinjo  | Date: 01/31/12 08:16:21 AM]
Reply
 
I do see the logic in much of what you are saying and think our inability to agree is because we are looking at this from different stand points. Forgive me if I'm wrong but I think that you are probably quite uncompromising as an enthusiast and not prepared to accept much less than best.

I'm a slightly different beast, more driven by budget, value for money and buying hardware which I can squeeze more out of either by overclocking or by looking at upgrade paths.

The latter is something that has made AMD very popular with enthusiasts many of whom value the ability of being able to upgrade their cpu on their current platform just as I'm describing.

The bottom line is, I understand, and you obviously also realise, that the majority of games played how gamers play them are gpu limited. There are the odd few that you highlighted that are cpu limited but they are by far in the minority.

Personally I'm quite happy to hang onto a platform in the knowledge that there will be little performance deficit on most of the games I play and a bit more on a few others.

I would only really consider renewing when my platform become incapable of playing what I want adequately rather than it was just giving me less than what newer hardware could.

The fact that an X79 platform would give me the option of upgrading the cpu from an i7 3820 to a 6-8 core IvyB-E without renewing the whole platform is just an added bonus.

I know when I come to do that in 2-3 years there will be plenty more capable cpu's available but I'm also sure that the IvyB-E will represent a worthwhile upgrade from the 3820 in terms of larger cache, clocks (particularly as I'd be getting a later stepping) and the numbers of cores which in spite of your doubts I think will be utilised in that time frame as more effort is being given to parallelism than ever before so you can't really compare previous progress in this field with what we can expect going forward.

As well as being a worthwhile upgrade I think it will also still be a viable cpu, at least as judged by my criteria, for several years subsequent to my upgrade.

0 0 [Posted by: technogiant  | Date: 01/31/12 10:30:06 AM]
Reply
4. 
Can anyone verify the release date of the i7-3820 as 13 February. I could find no info on Intel's website.
0 0 [Posted by: robert3892  | Date: 01/29/12 11:52:04 PM]
Reply
- collapse thread

 
It's true, trust me
1 0 [Posted by: Gavric  | Date: 01/30/12 03:35:28 AM]
Reply
5. 
show the post
0 3 [Posted by: londiste  | Date: 01/30/12 01:01:49 AM]
Reply
- collapse thread

 
2700k is better overclocker than 3820 and 3930k. We have tested all CPUs at its maximum stable frequency with same cooler.
2 0 [Posted by: Gavric  | Date: 01/30/12 03:34:36 AM]
Reply
 
i do understand that.

however, i would still argue that this way you lose one clear point of comparison (performance of different models on the same frequency) and gain a point of comparison that relies on the abilities of the specific processor samples. overclocking capability often varies a lot more than couple of hundred mhz within samples of the same processor model.
0 1 [Posted by: londiste  | Date: 01/30/12 04:52:04 AM]
Reply
 
You can always divide/multiply results proportionally to the clockspeed you want to achieve, and you will have decent approximation.
1 0 [Posted by: rrr  | Date: 01/30/12 11:18:38 AM]
Reply
6. 
What's the point of having a cpu without any software or o/s to run it?
0 0 [Posted by: tedstoy  | Date: 02/05/12 02:10:20 AM]
Reply
7. 
I have to say that I spent a lot of time trying to decide whether to go with Ivy Bridge or SNB/SNB-E, and I waited for months until the IVB processors were released to actually decide. I had the rest of my build all set out, but I couldn't buy the CPU/MB until the new architecture came out because I have a big fear of buyer's remorse. Needless to say, I was incredibly underwhelmed by the performance of Ivy Bridge, especially its overclocking ability, and yet I still waited until May '12 for the Maximus V but at that point it was nowhere, and I made my decision.
I decided to pass on the fully-tricked-out Case Labs TH10 with the 85mm Extended Ventilated Top, Pedestal, and a ton of other stuff to get the 3930K and Rampage IV Extreme. I came across the Switch 810 prior to this, and I loved the case and so with a $800 custom case not being in the budget anymore, I got my white Switch 810 instead of white TH10. I had planned to go with a custom water loop either way, and for my current single-GPU setup the 810 has more than enough room for rads, which I utilized by sticking an XSPC EX420 up top with 6x Bgears Blasters 140 1800rpm fans push/pull, an Alphacool UT60 240 rad in the bottom with 4x Koolance 120x25mm 2600rpm fans push/pull, 2 MCP35X pumps (waiting for the MCP35X2 Housing in white), EK Multioption Res X2 150 Advanced, Apogee HD CPU Block, and Heatkiller 680 Hole Edition GPU block and backplate. I have a MIPS full-copper Motherboard block, a Swiftech MCR320-XP rad, a set of RAM blocks (4x 2-DIMM sets), and an extra 680 block/backplate for when I get my second GPU.

The 2011 platform is, in my opinion, the only way to go if you want any degree of longevity. I don't upgrade my CPU/MB often, and when I do I go "all out", so the 2.5x number of PCI-E3.0 lanes, the 8xRAM DIMMs, tthe upupcoming IVB-E, and so forth make me confident that this platform will last. Even if it ends sooner than expected, the 3930K (and much less common and epeen-only 3960X) will remain strong performers for a very long time. One needs to simply look at the amount of people still running X58, most of whom seem to have a 920/930, and how well these processors have aged and then factor in the benefits of the X79 platform over X58 in terms of lifespan: X79 came out just in time to get the newest tech incorporated into it (the kind that takes long amounts of time to change), specifically USB3.0, SATA6G, and PCIe 3.0. Sure, DDR4 may come along, but I don't foresee any DDR4 surpassing DDR3 performance for at least a year or two after widespread release, aka 3-4 platforms from now.

Furthermore, Intel has no competition in the performance segment of the market, much less the Extreme performance end, so I don't see them pushing for faster and faster parts as it just doesn't make sense from a business standpoint. It costs a lot of money to design a new architecture, and it makes way more sense to make money off what you already have investments in. I think the progress for CPU tech is going to slow down dramatically until AMD makes another Phenom II or Athlon 64 FX (I still have my FX51 and Asus SK8V with an X800XT PE, btw).

Whether you play games or make them, watch video or produce it, or so forth, I am in strong favor ofSkt2011 and the 3930K. I definitely can't wait for the price of the EVGA GTX670 FTW to drop $50-100 so I can grab 2 more for 3-way SLI to go with the 27" H-IPS 2560x1440p display, which will be more than adequate for at least 2-3yrs, although I will likely just upgrade video cards until the next truly unique platform becomes available in 2015.

The cost of getting an 1155 board and chip actually ends up being significantly more than the 2011 platform if you look at how often you have to switch to an entirely new platform. 775 to 1156 to 1155 to 1150 just in 3 years, versus X58 to X79 lasting at least twice as long. It just makes sense.


Plus, I'd rather have a CPU cut from XEON wafers than from "mass-market" ones or have the chip filled with Solder instead of cheap TIM.

I realize many of the posts are old, but how many people still think that Ivy Bridge is better than Sandy, in terms of ACTUAL use (not LN2)and overclocking?
0 0 [Posted by: nleksan  | Date: 10/22/12 10:46:23 AM]
Reply

[1-7]

Back to the Article

Add your Comment


Our Sponsors

Crucial.com Western Digital G.Skill