Would heavily investing into X99 platform be a wise decition at this point?

[dave_the_nerd]

"Investing" is the wrong word, unless performance effects your income. (Say, you operate a render farm.)

If you can't do what you want to do with your system, then it's time to upgrade.

 

[pw257008]

"Investing" is the wrong word, unless performance effects your income. (Say, you operate a render farm.)

Hmm, I'd disagree. One can "invest" in future consumption. In the income function Income=Consumption+Investment, consumption is present consumption (ideally). If you buy a house to live in, that's an investment. Even if you're not working, just collecting social security and drawing down savings, you're investing in your future living standards. Even running a server farm is in the end just postponed consumption (whether consumption in goods/services or consumption of the comfort of having savings rather than living on the edge).

 

[BonzaiDuck]

Hmm, I'd disagree. One can "invest" in future consumption. In the income function Income=Consumption+Investment, consumption is present consumption (ideally). If you buy a house to live in, that's an investment. Even if you're not working, just collecting social security and drawing down savings, you're investing in your future living standards. Even running a server farm is in the end just postponed consumption (whether consumption in goods/services or consumption of the comfort of having savings rather than living on the edge).

+1 and :thumbsup: to that.

There is an economist currently at the "New School" formerly of U of Mass/Amherst(?) named Richard Wolfe, recently interviewed about 10 times by Bill Moyers on PBS. One of his assertions: universities really have "two economics departments" -- Economics, and Business Administration. The latter multi-discipline track tends to use the word more restrictively.

Suppose there is a commodity -- a durable good -- which you "need" more than want. It has a time-stream of benefits (returns -- not exclusively in dollars) over its lifespan. Your "investment" assures that your needs will be met at a certain reliable level -- perhaps long after the item has finished a straight-line depreciation schedule.

You can earn money with your computer -- I did with my various PCs over 20-some-years. You also save time, which can be translated into money, even if it is "leisure time." Then, there might be the improvements you make in managing your money, filing your tax-returns, or generally living with a modicum of discipline. I can think of myriad ways your life is improved, labor time decreased or various other facets which don't directly provide an "income stream" deposited in your bank account.

 

[coffeejunkee]

Yeah, there's no way any of us multitask enough to ever use 6 cores, or 12 workers...
Except DAILY here.. I want everything smooth & fast, and never have to worry about an under run while I'm doing 17 other things.. Just want my sh*t to work FAST.. Could care less about the electric bill.

Nice way of taking things out of context. Ofcourse there are people who benefit from 6 cores. But it's not the majority.

To really answer OP's question we would need to know what he uses his pc for. Since he didn't specify I'm assuming he doesn't have special needs like tons of 3D rendering or video-encoding.

So what his question boils down to, buy ultra high-end now and use it for 5 years. Or buy still very potent slightly less high-end, use it for 2,5 years, sell off rig, buy new stuff with even more features and use it for the next 2,5 years. I'd go with option 2.

 

[alyarb]

Swap out the 920 for a Xeon X5650 Hex core & ride it out for 2 more years.. And if you want USB3 & SATA III, add a card. Good cheap performance upgrade, while we see what Intel does, and how far prices fall.

X5650's are $70 and selling his 920 recoups most of that. A single add on card for USB 3 & SATA III is as low as $30. It's an inexpensive upgrade to carry him over & see what the future brings.

+1 of course



Did you ever find your elusive X5675 you've been looking for ?

http://forums.anandtech.com/showthread.php?t=2335636&highlight=l5639&page=65

X5675 are all over ebay for about $200. The MSRP was $1440.

I like this tactic and believe it may still be viable for X79 and X99 owners, which is one reason why it might not have been a terrible idea to buy into socket 2011 before the first tock.

People who are rocking a 3930k, or some other SNB-E will be able to drop in an IVB-EP, some years from now, when they are cheap like the westmere xeons of today.

The Asrock X79 Extreme6 for instance lists the E5 2690v2 and 2697v2 on its CPU support list. So does the ASUS X79 Pro and likely other desktop 2011 boards. Maybe 5 years from now these parts will also be on ebay for <$200 and people with X79's will also be mulling the alleged benefits of a migration similar to the OP's, when they know good and well there are 10 and 12 core fully compatible Ivy parts that are selling for less than a next-gen board would cost them.

It might even follow that Haswell-E builders may be rewarded too in 8-10 years with depreciated Broadwell-EP parts.

The 115x sockets can't do this for you. The trouble is that you have to weigh 12 old cores at 3 GHz vs 6 newer cores at 5 GHz or whatever Cannonlake-K alleges.

So if you are going long, you will spend more up front but your build stays relevant for about a decade. weirdly like consoles in that regard.

 

[mcbaes72]

X5675 are all over ebay for about $200. The MSRP was $1440.

I like this tactic and believe it may still be viable for X79 and X99 owners, which is one reason why it might not have been a terrible idea to buy into socket 2011 before the first tock.

People who are rocking a 3930k, or some other SNB-E will be able to drop in an IVB-EP, some years from now, when they are cheap like the westmere xeons of today.

The Asrock X79 Extreme6 for instance lists the E5 2690v2 and 2697v2 on its CPU support list. So does the ASUS X79 Pro and likely other desktop 2011 boards. Maybe 5 years from now these parts will also be on ebay for <$200 and people with X79's will also be mulling the alleged benefits of a migration similar to the OP's, when they know good and well there are 10 and 12 core fully compatible Ivy parts that are selling for less than a next-gen board would cost them.

It might even follow that Haswell-E builders may be rewarded too in 8-10 years with depreciated Broadwell-EP parts.

The 115x sockets can't do this for you. The trouble is that you have to weigh 12 old cores at 3 GHz vs 6 newer cores at 5 GHz or whatever Cannonlake-K alleges.

So if you are going long, you will spend more up front but your build stays relevant for about a decade. weirdly like consoles in that regard.

I've been researching another gaming/htpc type build. To get a used, high-end cpu with six cores for under $200 sounds good to me! Any mobo recommendations? TIA.

 

[biostud]

I don't think we will see revolutions in CPU design, that makes old CPU's totally obsolete, but just evolutionary designs with ~10% increase in speed/year. Performance/watt will probably increase much more per year.

Besides the actually new design/instructions in CPU's we also need software that can actually make good use of the better CPU's.

I would say a CPU bought today will probably be just as good an investment in 5 years as a CPU bought in 2 years will be in 7 years. I've just upgraded from my 4 year old i5-750 which could easily have had a year more of running since it was running @ 3.8GHz. I just had the upgrade itch. And I believe the 5820K will run fine for the next 5 years, for what I use my computer for.

 

[MongGrel]

X5675 are all over ebay for about $200. The MSRP was $1440.

I like this tactic and believe it may still be viable for X79 and X99 owners, which is one reason why it might not have been a terrible idea to buy into socket 2011 before the first tock.

People who are rocking a 3930k, or some other SNB-E will be able to drop in an IVB-EP, some years from now, when they are cheap like the westmere xeons of today.

The Asrock X79 Extreme6 for instance lists the E5 2690v2 and 2697v2 on its CPU support list. So does the ASUS X79 Pro and likely other desktop 2011 boards. Maybe 5 years from now these parts will also be on ebay for <$200 and people with X79's will also be mulling the alleged benefits of a migration similar to the OP's, when they know good and well there are 10 and 12 core fully compatible Ivy parts that are selling for less than a next-gen board would cost them.

It might even follow that Haswell-E builders may be rewarded too in 8-10 years with depreciated Broadwell-EP parts.

The 115x sockets can't do this for you. The trouble is that you have to weigh 12 old cores at 3 GHz vs 6 newer cores at 5 GHz or whatever Cannonlake-K alleges.

So if you are going long, you will spend more up front but your build stays relevant for about a decade. weirdly like consoles in that regard.

$200 too much compared to what a X5650 can do for under half that, is the point.

 

[DrMrLordX]

Can LGA2011-3 Xeons be overclocked? I'm guessing not.

 

[Mk pt]

ES xeon's can be overclocked.

Only 1366/x58 non-ES xeon's can be overclocked.

 

[DrMrLordX]

Thought so. There won't be a run on Haswell Xeons the way there has been on Westmere, just due to the absence of overclocking.

 

[Burpo]

ES xeon's can be overclocked.

Only 1366/x58 non-ES xeon's can be overclocked.

What? Engineering samples can definitely be overclocked..

 

[DrMrLordX]

Right, he's just saying that the last generation of retail/OEM (non-ES) Xeons to be overclockable was on LGA1366.

 

[Burpo]

http://www.chiploco.com/haswell-ep-xeon-e5-2699-v3-benchmarks-35799/

 

[gbeirn]

Why wouldn't BLCK overclocking work on the Xeons? On a X99 board?

 

[DrMrLordX]

http://www.chiploco.com/haswell-ep-xeon-e5-2699-v3-benchmarks-35799/

Okay, so you can exploit the turbo multi and 105 mhz multiplier for a little bump. That's still far short of the 4-4.5 ghz range that most of the consumer chips get. Of course, you do get many more cores if you really want them. How many end-users will really want that many cores, though?

I mean, I would, if I had the money for it, but I'm silly.

Why wouldn't BLCK overclocking work on the Xeons? On a X99 board?

bclk overclocking hasn't worked in awhile. You can hit maybe 105 mhz bclk before it craps out. That's the way it is on every Intel platform now.