Long term upgrade cycle P55 i5 / i7 vs X58 920

[cbn]

Originally posted by: RussianSensation


Futureproofing will hardly apply since I presume in 2 years you will also likely want PCI express 3, SATA3 and USB 3.0. So either platform is not futureproof.

(This is a serious question) How bottlenecked are we with SATA 2 and USB 2.0?

PCI-E 3.0 would certainly be a welcome addition to a mobo (considering how powerful Video cards are getting these days.)

 

[wand3r3r]

I think a X58 system with Windows 7 and a SSD could potentially last someone a really long time. When the prices of Core i9 CPUs begin to price drop (sometime in the Sandy Bridge era) then we are talking 6 cores @ 32nm with Hyperthreading. Six cores....six DIMM slots....two full speed x16 PCI-E 2.0 lanes.....there is a lot of potential here to accept increasingly powerful parts.

Good points. I am looking into purchasing an X58 MB + DDR3 + Cooler quite reasonably from an acquaintance so I believe I am going the X58 route if that goes through. Of course we'll have to see if the new features such as the mentioned USB3 and SATA3 will warrant a new MB anyway by the time 6 cores are affordable and faster then the i7 920.

Maybe an SSD drive would help the most? (But then you would need Windows 7 for the "trim" feature apparently.) But seriously I hear you. At the moment the only people who seem to benefit from quad cores are those using video editing software and gamers. (correct me if I am wrong). For everyone else dual core is plenty. In fact, I even wonder if power efficient laptops (benefiting from die shrinks rather than extra cores) will overtake desktops for this reason.

Again, good points... I have Windows 7 already so that's already solved. I also agree that at least currently I don't think I will utilize the 3rd and 4th core for my uses.

I'm anxiously awaiting FACTUAL pricing on the upcoming mainstream Intel X25 SSD's and the real details. Fudzilla specs revealed on X25 value SSD I HOPE it's larger then 40 GB and still affordable. Being it's slower then the X25M's as well as using smaller flash it should be cheaper but how much is the question. And is it so slow that it's not justifiably faster then the quicker HDs.

 

[wand3r3r]

Originally posted by: aigomorla

Originally posted by: wand3r3r

Cons:
- Quad-cores are EOL

which intel CEO did u hear this from directly?
Who told u there was no 32nm Quad offerings to come in the future?

And your not making any sense...
So you dont want to get a X58 because of its upgrade path.
Yet its a quadcore with HT which makes it almost like a 8 core.

But instead u want to look at the p55 and a dualcore so only in the future u can upgrade to what i described above?

Does this make any sense to you?

Originally posted by: wand3r3r

Yes, I meant the processors 920/40/50 at least are going to be discontinued according to the source. The socket will be kept and should support the 6-cores anyways but I haven't heard anything about i7 930 or 960s. That would be interesting to know, I think this is the first I have heard of that. Does anyone have any information about them?

the 960 is scheduled to release.
And a rumored 930 as well.

Also the gulftowns dont come out til april, so yeah 32nm quad offerings WONT show up on the sheets YOUR looking at.

Once again.. where the heck are you pulling all this info from?

I have only seen roadmaps which have dual cores for 1156, and 6-cores for 1366. As I mentioned previously I believe I would benefit more from fewer cores but faster cores. Of course I haven't seen the concrete Intel roadmaps. Come to think of it, where are the 32nm quads? I haven't seen them in any roadmap but it seems they would still produce them.

About the upgrade path, I was only thinking that the X58 doesn't have any KNOWN upcoming 32nm processors so why spend extra $? That was just an opinion.

 

[wand3r3r]

Originally posted by: Just learning

Originally posted by: RussianSensation


Futureproofing will hardly apply since I presume in 2 years you will also likely want PCI express 3, SATA3 and USB 3.0. So either platform is not futureproof.

(This is a serious question) How bottlenecked are we with SATA 2 and USB 2.0?

PCI-E 3.0 would certainly be a welcome addition to a mobo (considering how powerful Video cards are getting these days.)

Aren't the top of the line SSD's approaching the bottleneck or was it only SATA1 which was about saturated? I have seen something somewhere but don't remember for sure about that. I think it was SSDs may reach the saturation speed of SATAII. As for USB3, I would like it for faster flash drives but it is not worth paying extra at this point for me.

 

[RussianSensation]

Originally posted by: wand3r3r

Aren't the top of the line SSD's approaching the bottleneck or was it only SATA1 which was about saturated? I have seen something somewhere but don't remember for sure about that. I think it was SSDs may reach the saturation speed of SATAII. As for USB3, I would like it for faster flash drives but it is not worth paying extra at this point for me.

USB 2.0 is a major bottleneck already!
USB 2.0 vs. eSATA - OCZ Throttle
Right now the faster USB drives have a huge premium. But with progress USB 3.0 should make faster technology more accessible at reasonable prices.

As far as SATA 3.0 is concerned, SSDs are closing in on 250-260MB/sec. While right now it's not a bottleneck, I would like to think in 2 years from now we will have SSDs that are much much faster. Basically when Core i7 860 / 920 become relatively midrange (2 years), SATA 3 drives may be exceeding 300MB/sec (I hope).

PCIe 3.0 is likely to be the least beneficial in the meantime.

My point is we have had SATA II, USB 2 and PCIe 2 for a long time now. So we can expect another serious wave of performance innovation within 1-2 years. At which point I am not sure if the argument of 'futureproofing' applies to current platforms.

 

[Jumpem]

Originally posted by: RussianSensation
My point is we have had SATA II, USB 2 and PCIe 2 for a long time now. So we can expect another serious wave of performance innovation within 1-2 years. At which point I am not sure if the argument of 'futureproofing' applies to current platforms.

Exactly. Everyone going on about "futureproofing" and "upgrade paths" will drop whatever they have (s1156 or s1366) for Sandy Bridge. It's a moot point really.

 

[cbn]

Originally posted by: RussianSensation

Originally posted by: wand3r3r

Aren't the top of the line SSD's approaching the bottleneck or was it only SATA1 which was about saturated? I have seen something somewhere but don't remember for sure about that. I think it was SSDs may reach the saturation speed of SATAII. As for USB3, I would like it for faster flash drives but it is not worth paying extra at this point for me.

USB 2.0 is a major bottleneck already!
USB 2.0 vs. eSATA - OCZ Throttle
Right now the faster USB drives have a huge premium. But with progress USB 3.0 should make faster technology more accessible at reasonable prices.

As far as SATA 3.0 is concerned, SSDs are closing in on 250-260MB/sec. While right now it's not a bottleneck, I would like to think in 2 years from now we will have SSDs that are much much faster. Basically when Core i7 860 / 920 become relatively midrange (2 years), SATA 3 drives may be exceeding 300MB/sec (I hope).

PCIe 3.0 is likely to be the least beneficial in the meantime.

My point is we have had SATA II, USB 2 and PCIe 2 for a long time now. So we can expect another serious wave of performance innovation within 1-2 years. At which point I am not sure if the argument of 'futureproofing' applies to current platforms.

Thanks for the info.

So measured IO is what I need to look at if I am worried about a SSD being too fast for a particular SATA port?

 

[imported_Lothar]

Originally posted by: Jumpem

Originally posted by: RussianSensation
My point is we have had SATA II, USB 2 and PCIe 2 for a long time now. So we can expect another serious wave of performance innovation within 1-2 years. At which point I am not sure if the argument of 'futureproofing' applies to current platforms.

Exactly. Everyone going on about "futureproofing" and "upgrade paths" will drop whatever they have (s1156 or s1366) for Sandy Bridge. It's a moot point really.

Not me.

 

[cbn]

Originally posted by: Jumpem

Originally posted by: RussianSensation
My point is we have had SATA II, USB 2 and PCIe 2 for a long time now. So we can expect another serious wave of performance innovation within 1-2 years. At which point I am not sure if the argument of 'futureproofing' applies to current platforms.

Exactly. Everyone going on about "futureproofing" and "upgrade paths" will drop whatever they have (s1156 or s1366) for Sandy Bridge. It's a moot point really.

Yeah but lots of people with existing 45nm Core 2 are still happy even with 45nm Nehalem out.

The issue of the SATA 2/3 ports with faster SSDs sounds more interesting to me than Sandy Bridge.

 

[wand3r3r]

Excellent posts here. You guys have brought many relevant points which have basically helped me determine that I was overthinking the 'upgradability'...

As far as SATA 3.0 is concerned, SSDs are closing in on 250-260MB/sec. While right now it's not a bottleneck, I would like to think in 2 years from now we will have SSDs that are much much faster. Basically when Core i7 860 / 920 become relatively midrange (2 years), SATA 3 drives may be exceeding 300MB/sec (I hope).

Just to break down what RussianSensation mentioned:
SATAII has a max speed of 3 Gb/s = 375 MB/s. If the drives are achieving some in the mid to upper 200 MB/s they are encroaching upon saturating the SATAII bus.
SATAI - 1.5 Gb/s = 187.5 MB/s. They are already far beyond saturating the SATA-I bus and the SSD would be bottlenecked in a SATA-I system.

Those numbers are only theoretical and my guess would be that the performance would likely be reduced even as you near the saturation point. I also would like to think that SATAII will be insufficient in a year or two because that means the SSD's will really perform!

 

[lopri]

'Long term upgrade cycles' and 'Intel' don't go hand-in-hand..

 

[cbn]

Originally posted by: wand3r3r
Excellent posts here. You guys have brought many relevant points which have basically helped me determine that I was overthinking the 'upgradability'...

As far as SATA 3.0 is concerned, SSDs are closing in on 250-260MB/sec. While right now it's not a bottleneck, I would like to think in 2 years from now we will have SSDs that are much much faster. Basically when Core i7 860 / 920 become relatively midrange (2 years), SATA 3 drives may be exceeding 300MB/sec (I hope).

Just to break down what RussianSensation mentioned:
SATAII has a max speed of 3 Gb/s = 375 MB/s. If the drives are achieving some in the mid to upper 200 MB/s they are encroaching upon saturating the SATAII bus.
SATAI - 1.5 Gb/s = 187.5 MB/s. They are already far beyond saturating the SATA-I bus and the SSD would be bottlenecked in a SATA-I system.

Those numbers are only theoretical and my guess would be that the performance would likely be reduced even as you near the saturation point. I also would like to think that SATAII will be insufficient in a year or two because that means the SSD's will really perform!

Apparently ASRock is making a SATA 3 card for PCI-E x1 slots.

So SATA2 mobos should still be able to make use of faster SSDs in the future.

 

[aigomorla]

Originally posted by: wand3r3r
I was only thinking that the X58 doesn't have any KNOWN upcoming 32nm processors so why spend extra $? That was just an opinion.

so what kind of upgrades were u looking for?

the floor model is a 4 core 8 thread monster.

There is 2 versions of this chip. C0/C1 and the infamous D0.

32nm gulftowns are coming in April.

Your not going to hear about the 32nm 4cores until the 6 cores have debuted.

However right now theres so much red tape on gulftown, i doubt ur gonna really see one, minus leaks until middle november/december.

 

[wand3r3r]

Well, I was just sitting on the fence because there are 2 sockets with not as much performance difference as is the price difference. This is what I was hoping to gather opinions and information for, to help make the decision.

Your not going to hear about the 32nm 4cores until the 6 cores have debuted.

My question is which socket are the 32nm quads going to be for? This information is not (widely) available and would help make the decision easier. Of course I can speculate they will continue X58 quads but that's not concrete.

 

[aigomorla]

uhhhh if i told u i got gulftown to work on my classified with amazing results...
would that make you happy?

Gulftown 32nm Hexcore = working on a X58 chipset.

So i dont see why 32nm i7 would not work on a X58 chipset.

I dont think they are intending to EOL the X58 for a while either.

 

[RussianSensation]

Originally posted by: wand3r3r

SATAII has a max speed of 3 Gb/s = 375 MB/s. If the drives are achieving some in the mid to upper 200 MB/s they are encroaching upon saturating the SATAII bus.

Interestingly enough, this SATA Report, Page 8 outlines that the maximum throughput bandwidth of SATA II is actually 270 MB/sec. As far as I remember SATA II interface is actually rated at a maximum of 300 MB/s, and not 375 MB/s. In this case, current SSD hard drives are getting dangerously close to saturating the current standard.

 

[wand3r3r]

Originally posted by: RussianSensation

Originally posted by: wand3r3r

SATAII has a max speed of 3 Gb/s = 375 MB/s. If the drives are achieving some in the mid to upper 200 MB/s they are encroaching upon saturating the SATAII bus.

Interestingly enough, this SATA Report, Page 8 outlines that the maximum throughput bandwidth of SATA II is actually 270 MB/sec. As far as I remember SATA II interface is actually rated at a maximum of 300 MB/s, and not 375 MB/s. In this case, current SSD hard drives are getting dangerously close to saturating the current standard.

Good source. Yeah it does say "SSD's ... utilize all of the available 270 MB/s bandwidth of SATA 3Gb/s". So it's not a true 3Gb/s bus but probably uses some bandwidth for control or else just can't quite reach the 'theoretical' limit. Basically that means that SATA3 is already basically useful. If the faster SSD's saturate the SATAII bus already then it will only take a small iteration and they will be bottlenecked.

 

[wand3r3r]

I was able to pick up an Asus P6T SE, Noctua NH-C12P, OCZ CL8 DDR3 3x2GB 1600 MHz, and an Antec Earthwatts 500 psu. I will reuse my 4830 OC. Now I just need to get an i7 920.

I am questioning if I will be able to overclock with the Earthwatts psu? It's not very powerful but I know at stock speeds that system will not utilize all the power. Does anyone have a similar setup and does it overclock fine? I am thinking of overclocking without modifying voltages and some sites seem to get higher speeds then others but my goal will be 3.5-3.8 GHz or so.

An interesting note on the Noctua NH-C12P cpu cooler. The Asus P6T SE has LGA 775 cooler holes as well as 1366 cooler mounting holes. However the mounting kit seems a little bit awkward and looking on Noctua.at it says the P6T SE is supported with a 1366 Mounting kit which they will ship you for free, provided you have a receipt for the 1366 MB, and Fan. This is applicable to all Noctua owners! It's a pretty good feature for all Noctua owners to keep the fan with their next build. (note the up to 2 week shipping however)
FREE Noctua 1366 Mounting kit (Fill in form)

The Noctua NH-C12P was able to fit in the Asus P6T SE motherboard holes but it seems like you would need spacers on the back of the MB or else it feels like the backplate bracket would bend.

 

[aigomorla]

you'll be fine as long as you dont go sli or xfire happy.

 

[cbn]

Originally posted by: wand3r3r
The Asus P6T SE has LGA 775 cooler holes as well as 1366 cooler mounting holes.

That is a pretty nice feature. I will have to keep my eye out for that in the future. It is nice to be able to carryover the CPU cooler.

P.S. Speaking of SATA3, it looks like Gigabyte is already preparing this for P55-->http://www.tweaktown.com/news/...x_usb_power/index.html

 

[wand3r3r]

From the Gigabyte P55A article

GIGABYTE told us that they saw up to a 30% increase in performance when going from a SATA2 to SATA 3 hard drive under HD Tach, measuring both burst and read speeds.

The also article mentions they are going to preview Gigabytes X58 Extreme 2 motherboard. Venturing a guess it seems like it will include USB3, SATA3 as well (Based on the fact they are previewing a new P55 with those new features and advertising a preview next week for the same manufacturers updated X58 MB)!

I am not sure how the LGA 775 coolers actually fit on the P6T boards as I got a 1366 mounting kit anyway.

 

[Suspicious-Teach8788]

This is the same fear I have but time has shown that I have never been able to futureproof ANYTHING.

Why? Because what's your upgrade cycle? If it's 1 year, 2 year you might be ok. But even if got LGA775 for all 4 years, would your original LGA775 board have worked with today's C2Quads? I doubt it. Did they even have PCI-e back when LGA775 first came out? SATA2?

DDR3 is still maturing, but if your upgrade cycle is 3-4 years, you might consider that by the time you want to seriously upgrade, DDR4 might be inching there already. Once you hit either a socket change, memory change, or say PCI-e change, you need to throw out your mobo. With a new mobo comes a new cpu and new ram, etc etc etc. So in the end you have to build a new system anyway. The only things you keep around are HDDs and optical drives.

My minimum upgrade right now is a CPU+MOBO+RAM. That spells close to $400 for an i5 system, and certainly more for an i7.

 

[Makaveli]

I was just in this same situation and went the 920 D0 route.

after I priced together 3 builds

1. AMD
2. Lynnfield
3. Bloomfield

The price difference was option 3 cost $300 more than option 1 and $200 more than option 2. For so little of a price difference 1366 was the best option. Not to mention there are better Air coolers available for this socket and it general takes less voltage to push a 920 D0 to 4Ghz than a lynnfield chip.

As for the usb 3.0 Sata III issue its called a add in card. Next thing I believe we will see cheaper 6 core gulftowns on the 1366 platform next year and just not an extreme edition. Alot of people keep taking about this to scare people away from this socket.

I upgraded from a Opteron 170 with 4GB of ram, and even at stock speed the 920 is so much faster its not even funny.

Also I didn't have to spend $400-$500 on a 1366 board, I picked up an Asus P6T Deluxe V2 for $300 where as the lynnfield board I wanted was $250 Cdn prices!

 

[RussianSensation]

Makaveli those are some steep prices man. To say that $200/$300 more for 1366 platform may not be a big deal until you consider that's a free Intel Gen 2 SSD, or another 5850 graphics card.

The biggest problem on the 1156 socket is that the memory DIMM slots are very close to the cpu heatsink. But with Megahalems 1156, that's about all of the air cooling you need After 4.2+ ghz, water is probably the only safe way.