Would upgrading a I7-930 to a Xeon X5680 be worth it?

[ClockHound]

The real issue with X58 is not just the CPU, no SATA 3, no PCI-E3, no USB 3.

No SATA 3, no PCI-E3, no USB 3. No worries.

SATA3 is overrated - SATA2 is fine for spindles and way more than enough bandwidth for the key feature of SSDs - random i/o. Anyhow, most premium X58 boards have SATA3, provided by an under performing 3rd party chipset. So that's a feature win! (Sort of).

PCI-3...2-3% more performance for the 1%.

USB3? For under $30 I can get a USB 3.1 pci card. And connect my super sharp smartphone from 2012. Anyhow, most premium X58 boards have USB3, provided by an under performing 3rd party chipset. So, that's another...uh...win!

Lack of NVme support is possibly the only feature that would move my lazy ass off the X58 farm. If there was a strong value proposition - which will only happen when NVme is deprecated for the latest nvMMMMMMram passing standard.

For enthusiasts, OC'd 56xx Xeons on X58 are the last bright flicker in the fading golden age of personal desktop computing. An anomaly that Intel will not allow to happen ever again. The grand children will never believe the stories about buying Intel hex-core server chips at 5-10 cents on the dollar and tweaking them to out-perform quad cores 5 generations newer.

The hunger to upgrade from a hotrod X58/Xeon is more about fashion than function. Nothing wrong about wanting to be fashionable. Personally, I enjoy being fashionably...

...late. ;-)

 

[Burpo]

Well said CH.. Sabertooth has USB 3.0 anyway

 

[Headfoot]

$100 to go from 4 core 45nm to 6-core on Intels excellent 32nm? Definitely.

 

[mysticjbyrd]

$100 to go from 4 core 45nm to 6-core on Intels excellent 32nm? Definitely.

For what? 2 fps in some games? No, not worth at all.

 

[Makaveli]

For what? 2 fps in some games? No, not worth at all.

You do know that some of us do more than just play games on our computers.

Jesus you kids!

 

[scannall]

For what? 2 fps in some games? No, not worth at all.

What games would those be exactly? I think finding examples will prove pretty difficult for ya.

 

[readers]

No SATA 3, no PCI-E3, no USB 3. No worries.

SATA3 is overrated - SATA2 is fine for spindles and way more than enough bandwidth for the key feature of SSDs - random i/o. Anyhow, most premium X58 boards have SATA3, provided by an under performing 3rd party chipset. So that's a feature win! (Sort of).

PCI-3...2-3% more performance for the 1%.

USB3? For under $30 I can get a USB 3.1 pci card. And connect my super sharp smartphone from 2012. Anyhow, most premium X58 boards have USB3, provided by an under performing 3rd party chipset. So, that's another...uh...win!

Lack of NVme support is possibly the only feature that would move my lazy ass off the X58 farm. If there was a strong value proposition - which will only happen when NVme is deprecated for the latest nvMMMMMMram passing standard.

For enthusiasts, OC'd 56xx Xeons on X58 are the last bright flicker in the fading golden age of personal desktop computing. An anomaly that Intel will not allow to happen ever again. The grand children will never believe the stories about buying Intel hex-core server chips at 5-10 cents on the dollar and tweaking them to out-perform quad cores 5 generations newer.

The hunger to upgrade from a hotrod X58/Xeon is more about fashion than function. Nothing wrong about wanting to be fashionable. Personally, I enjoy being fashionably...

...late. ;-)

Loads of non sense, when X58 came out the standard for Sata 6.0 gbps and USB 3.0 wasn't even finalized, I don't care about how premium your board is, BS won't get you a feature that simply did not exist at the time. Some later X58 boards have them, not premium or not, just like top of the line X99 came out in 2014 did not have 3.1 but a cheap one today might have it.

Also stop BSing about how Sata 6gbps doesn't make a huge difference, I was using SSD with 3gbps sata for 2 years and upgrade to 6 gbps is huge.

"5-10 cents on the dollar "

Might want to check ebay price on them now then compare to a 6700k or 5820k, cheaper, yes, but 5-10 cents on the dollar, no. And they do not outperform either, stock to stock, oc to oc. If you want to compare stock to OC... I don't know the result, but I know it's pointless.

 

[pooptastic]

The things that are going to get me to upgrade to anything new are likely that new M.2 or U.2? SSD slot thingie that's out.

Love SSDs, my 850 pro is the bees knees, I can't imagine an even faster drive than that.

 

[readers]

The things that are going to get me to upgrade to anything new are likely that new M.2 or U.2? SSD slot thingie that's out.

Love SSDs, my 850 pro is the bees knees, I can't imagine an even faster drive than that.

U2, 3.1, PCI-E 3.0, which is big one if you do SLI, but otherwise not too big.

Lastly,

http://cpu.userbenchmark.com/Compare/Intel-Core-i7-X-980-vs-Intel-Core-i7-5820K/m1822vs2579

Sure, X58 OC very well, but X99 OC very well too.

If you don't want to OC

http://cpu.userbenchmark.com/Compare/Intel-Core-i7-X-980-vs-Intel-Core-i7-6700K/m1822vs3502

 

[Makaveli]

Also stop BSing about how Sata 6gbps doesn't make a huge difference, I was using SSD with 3gbps sata for 2 years and upgrade to 6 gbps is huge.

The difference is only noticeable with sequential file transfers for the most part.

For the windows desktop i've had it both on SATA 2 & 3 in another system for a friend and without telling him which drive was which he couldn't tell the difference for normal desktop usage.

Its the same argument being used now for PCIe NVME drives vs SATA III drives.

In normal desktop usage you won't feel the difference from the SATA drive that reads at 500 Mb/sec to the NVME drive that reads at 1500 Mb/sec.

The difference between PCI 2.0 and PCI 3.0 is also small to non existent on single GPU's setups.

So he does make some valid points.

On my system having ESATA access has not made me miss USB 3.0 at all.

Granted if I was building a system now of course I would go with something brand new, but if his system still meets his needs a drop cpu replacement is a no brainer right now for $100.

 

[Burpo]

If you want to compare stock to OC... I don't know the result, but I know it's pointless.

Your post is pointless

Obvious that you have no X58 (Hexcore) experience to even offer advice..

 

[readers]

Your post is pointless

Obvious that you have no X58 experience to even offer advice..

LOL, week 1 920, ran it at 3.8G 24/7 for 6 years. 5930k now, night and day difference.

 

[Exzeph]

So much like you I was on a i7 930 (Intel Gen -3 Represent!!), and then I went down this same path and found out nearly years ago that I could get a die-shrink and 2 more cores and efficiency by staying on the same X58 chipset and blowing like ~90-130 bucks on ebay.

So yeah, I went with the Xeon 5670 over any of the higher stock clocked versions for the exact reasons Burpo iterated earlier on. I'm not crunching any kind workload, so I run a really low-level 4 Ghz overclock just to keep it peppy for modern games and not too hot for normal use. It idles pretty well at 26-32c with a Megahelams tower cooler.

And well, I haven't been convinced that for my uses (basically just gaming and a tiny little bit of content creation), that any of the mainstream chips (with only 4 cores) can really provide any kind of real-world end user benefit over what I run, at least within the usage cases that I have.

It seems to me that the trend in gaming is going towards more thread utilization, in the near future so I'm wondering if it might be a bit embarrassing in the near future when DX12 games seem to do pretty well on 12 Threaded x58 Xeons to the point of slightly embarrassing these new-fangled 8 threaded or 4 threaded (i5) Skylakes.

I mean sure on a benchmarks I am not going to put up the ultra high numbers, but okay, I don't live to bench.

Frankly the decision to upgrade is easy. Even if you want to build a brand new PC -- which would make sense for anyone with any kind of multi-thread workload and who wants to get in on those 7 generations of incremental IPC improvements. --why not spend the minor $ to keep a relevant second machine that could simply be used for casual use and gaming.

 

[YBS1]

LOL, week 1 920, ran it at 3.8G 24/7 for 6 years. 5930k now, night and day difference.

A 920@3.8 versus an X56xx@4.5 is night and day too, probably about 50-60% of the jump to a 5930k.

 

[RussianSensation]

LOL, week 1 920, ran it at 3.8G 24/7 for 6 years. 5930k now, night and day difference.

Ok but let's say he sells his existing 930 and gets a $100 stop-gap Xeon? That will allow him to coast until Broadwell-E or maybe until Skylake-E. If he has a limited budget between his CPU/GPU upgrade choices, it makes way more sense to get the Xeon for $100 while selling the 930 and investing the rest into a GPU upgrade. If his existing GPU is becoming outdated (let's say GTX680/770/780/7970Ghz), then blowing $500-700 on a Skylake/X99 platform will do almost nothing for his gaming experience vs. a 4.5Ghz 6-core Xeon from that era.

USB 3.1 -- nice marketing gimmick of 2015. Until there are many USB 3.1 flash drives at good prices or way faster mechanical HDDs that can actually tap into USB 3.1 speeds, it's a marketing gimmick.

SATA 3 -- yes, this is good but only for sequential reads and writes but for every day operations, random access is king and in this area SATA 2 is enough for him to feel the difference.

PCIe 3.0 -- 6% difference at 1080P on a 980. Unless the OP has a 980, this isn't even a factor. Besides, 1-2 settings in any game or cranking AA and that 6% advantage will get wiped out into GPU-limited land in modern titles.

You are also low-balling the performance of a 32nm 6-core Xeon OC. It performs extremely well for its age.

 

[RussianSensation]

Right now jumping on the X99 platform makes no sense considering 5xxx CPUs have been around since August of 2014 and prices on them haven't fallen. If going X99, it's prudent to wait for BW-E.

For someone who has kept his system for so long, there is no way imo that an i5-6600K is a good recommendation because in some games it gets its ass handed by the 2600K.
http://www.sweclockers.com/test/20862-intel-core-i7-6700k-och-i5-6600k-skylake/14#content

This is one of those cases where the OP has very little to lose but imo spending $500 on DDR4, Z170 and i5-6600K isn't as stellar as it sounds. For productivity an i5 is mediocre, while it's always nice to ignore the benefits of HT for i5 users to justify their purchase, the reality can be different. If someone is spending $400+ on a platform upgrade for the next 5 years, might as well not take chances for the next 2-3 GPU upgrades and go all the way to the i7 6700K.

That means in reality for someone who wants to keep his system just as long as his 930, it would be better to get the 6700K but that keeps raising the price of his upgrade even more -- $$$ that he'd be better off using to upgrade his graphics card I am sure. If the OP could do an i7-6700K platform upgrade + a GTX980Ti upgrade, I am sure he wouldn't be contemplating keeping his X58.

Finally, since the demand for 32nm Xeons is high, if the performance/overclocking doesn't meet his needs, he can always sell it back for $100 and go Broadwell-E or Skylake/SKL-E. That means the Xeon stop-gap option is extremely cost effective from a cost of ownership point of view. The same absolutely cannot be said of 5820K/5930 as of right now as their resale value will drop a lot more once BW-E release vs. the resale value of a Xeon.

However, it's always good to consider several options. One other possibility is putting up the entire X58+930+DDR3 memory up for sale. With that in mind, the full platform swap to an i7-6700K wouldn't be $600; so you do have a point

 

[cytg111]

1. NVMe++, thats not happening on x58 IIRC. That would be important to me.
2. Nehalem is just behind the 'flattening' of the clock/ipc curve of sandybridge and onwards. That means, to me, that a simple one-two step cost/benefit analysis is sub-par.. think of all the ST performance you're missing out on over the next 5-6-7 years(cause that goalpost wont be moving..)

 

[YBS1]

2. Nehalem is just behind the 'flattening' of the clock/ipc curve of sandybridge and onwards. That means, to me, that a simple one-two step cost/benefit analysis is sub-par.. think of all the ST performance you're missing out on over the next 5-6-7 years(cause that goalpost wont be moving..)

It's a simple sub $100 drop in upgrade. If someone is looking for it to carry them 5-7 years into the future they are being more than optimistic. They shouldn't be looking at the 6600K either at that point versus the price difference to the 6700K, to that end they should even consider throwing the 5820K in the mix.

 

[cytg111]

Hell I would problary do the upgrade for the sole reason of not having to install a new os and all my * all over again. Still, seeing how ipc/clock scaling is dead in the water, there is a good argument to get to the top of the game now and stay there for the next ~10 years?

 

[MrTeal]

Hell I would problary do the upgrade for the sole reason of not having to install a new os and all my * all over again. Still, seeing how ipc/clock scaling is dead in the water, there is a good argument to get to the top of the game now and stay there for the next ~10 years?

If you're building a new system now, it makes a lot of sense to go higher end and higher core count. I just built with a 5930k system to replace my 2500k since I wanted to do a big water rig and might as well toss some good hardware in there if you're going to put in all that work. I'd expect it to last quite awhile, but it was also over a grand for CPU, MB and new RAM. That's an entirely different ballpark than buying a $1500 Xeon for under $100 and tossing it in there. His X58 board has many of the modern features you'd want like USB 3 and SATA3, so unless he really needs USB3.1 or wants to go Tri-SLI it's hard to argue against an upgrade that will cost him a net of $50 and an hour of his time.

 

[pooptastic]

It's my 2nd machine and I think at this price point it would be a nice upgrade for that system.

My main station is a 3930k, which I absolutely love. Aren't those a rebranded Xeon themselves? I like how I can keep Firefox and Chrome and a ton of other crap open and just launch a game with 0 observable degradation in performance.

Maybe it's the extra cache, maybe it's the 2 extra cores. Maybe it's a placebo and all in my head, but after reading here I could swap out a Xeon hex core into my older i7-930 system so cheaply, I'm excited to give it a shot.

That older system has a 6970 still. My 3930k has a lowly 670 gtx, both of which could use replacing, but I'm patient and would rather wait until next year as I feel the Pascal launch will be a large generational leap in graphics performance.

 

[RussianSensation]

1. NVMe++, thats not happening on x58 IIRC. That would be important to me.

But you aren't the OP so we should put ourselves into his shoes.

-> 400-512GB NVMe drive is $320-400 US

Do you think someone contemplating a CPU vs. GPU upgrade is going to be caring about spending $300-400 on an 400GB PCIe drive? No.

Still, seeing how ipc/clock scaling is dead in the water, there is a good argument to get to the top of the game now and stay there for the next ~10 years?

Wouldn't that mean it's then better to get a 6-core Broadwell-E or wait 1.5 years to 6-core Skylake-E?

2. Nehalem is just behind the 'flattening' of the clock/ipc curve of sandybridge and onwards. That means, to me, that a simple one-two step cost/benefit analysis is sub-par.. think of all the ST performance you're missing out on over the next 5-6-7 years(cause that goalpost wont be moving..)

Why would he need to keep the Xeon for 5-7 years? If you had read the thread we are saying to use it as a stop-gap solution perhaps to get him to Broadwell-E or Skylake-E in 2017. If he were to just upgrade to an i7 6700K platform, he would have no money left for his GPU upgrade and his gaming experience wouldn't improve that much vs. dropping in a 6-core Xeon + 4.4Ghz overclock + a new GPU. It's about figuring out the best way to spend the money when there are limited budget constraints.

As I mentioned earlier, with a 4.4-4.6Ghz 6-core Xeon, he would be more GPU limited than CPU limited if his GPU is GTX780 or below.

Based on his last reply, his GPUs are well below GTX780 level.

That older system has a 6970 still. My 3930k has a lowly 670 gtx, both of which could use replacing, but I'm patient and would rather wait until next year as I feel the Pascal launch will be a large generational leap in graphics performance.

Ya, in that case the choice is even easier. If i7 930 rig has an HD6970 in there, going all in on a 6700K platform upgrade while leaving even less $ for 2 GPU upgrades doesn't sound like a good strategy, unless you can afford it.

Perhaps something to think about is you can sell the i7 930 platform, make 3930K your 2nd system and get a 6700K/BW-E as your primary. I guess you have a lot of options but both of your GPUs are the primary bottlenecks, especially the 6970.

 

[cytg111]

Do you think someone contemplating a CPU vs. GPU upgrade is going to be caring about spending $300-400 on an 400GB PCIe drive? No.

I do so get where you are coming from, still, for the ~10% fps it's gonna net him.. not worth it imo. even for 50$, for the sheer hazzle of 1. anadtech forums, 2. ebay'ing(with the uncertainty that follows), 3. upgrading rig and hope everything is stable. Nah, wouldnt touch it.

 

[Makaveli]

I do so get where you are coming from, still, for the ~10% fps it's gonna net him.. not worth it imo. even for 50$, for the sheer hazzle of 1. anadtech forums, 2. ebay'ing(with the uncertainty that follows), 3. upgrading rig and hope everything is stable. Nah, wouldnt touch it.

There is no number 3.

If his system is stable with a 930 it will be stable with a drop in replacement Xeon.

When it comes to overclocking he will have to find the sweet spot for his chip like he did for the 930. But I disagree with saying the dropin replacement will cause stability issues cause it should be compared at stock and overclocking should not be added to that because that is something done after the fact on both chips. He has to find stable overclock regardless of what chip is in the socket.

 

[MrTeal]

I do so get where you are coming from, still, for the ~10% fps it's gonna net him.. not worth it imo. even for 50$, for the sheer hazzle of 1. anadtech forums, 2. ebay'ing(with the uncertainty that follows), 3. upgrading rig and hope everything is stable. Nah, wouldnt touch it.

He's already got a SB-E system with a 670 in it that would be a much better gaming rig, so baring him saying this upgrade is for gaming purposes I'm not sure it's safe to discount it just because it won't net a massive FPS increase. There's a lot other things that would see a big benefit from 50% more cores, especially if they're clocked higher.