Core i7 7700k benchmark

[patrickjp93]

52 minutes ago, VagabondWraith said:

Does the chipset affect performance? Like, would I be better off with a z270 board and 7700K over a z170 and 7700k?  Or is the chipset different from actual CPU performance? Jw 

If you want multiple USB 3.1 devices in use at once, then having the new chipset will be helpful. If you want the performance of Optane SSDs, then you will want Z270. Otherwise, all the performance gain is in the CPU.

[Hayabusa1989]

2 hours ago, patrickjp93 said:

Coffee Lake will be bringing that.

Is that a brilliant joke or are you being serious ?

[patrickjp93]

3 minutes ago, Hayabusa1989 said:

Is that a brilliant joke or are you being serious ?

I'm completely serious. It's been on Intel's official roadmap for 2 months now. It's launching at the same time as Cannonlake and puts 6 cores on the mainstream socket.

[Hayabusa1989]

3 minutes ago, Hayabusa1989 said:

Is that a brilliant joke or are you being serious ?

 

26 minutes ago, patrickjp93 said:

If you want multiple USB 3.1 devices in use at once, then having the new chipset will be helpful. If you want the performance of Optane SSDs, then you will want Z270. Otherwise, all the performance gain is in the CPU.

Oh wow, Fair Enough - CannonLake might just be amazing then

[Bleedingyamato]

7 hours ago, Kavawuvi said:

USB 3.1 and Thunderbolt 3 already exist, but require an additional component on the motherboard. It'll be supported from the chipset, meaning you may see it on a lot more motherboards, including cheaper ones.

That would be nice.  Faster options for copying data on more motherboards would be helpful.

 

Though we also need SSDs to get larger and cheaper so that extra speed could be fully utilized.  Currently I think USB 3.1 is faster than what a 7200rpm HDD can handle right?

[Bleedingyamato]

4 hours ago, patrickjp93 said:

If you want multiple USB 3.1 devices in use at once, then having the new chipset will be helpful. If you want the performance of Optane SSDs, then you will want Z270. Otherwise, all the performance gain is in the CPU.

What's an optane SSD?  

[SnowyMus]

52 minutes ago, Bleedingyamato said:

That would be nice.  Faster options for copying data on more motherboards would be helpful.

 

Though we also need SSDs to get larger and cheaper so that extra speed could be fully utilized.  Currently I think USB 3.1 is faster than what a 7200rpm HDD can handle right?

A 7200 RPM hard drive cannot come close to saturating a SATA 6 Gbps connection without RAID. In fact, they have a hard time breaking SATA 3 Gbps. USB 3.0 operates at 5 Gbps and USB 3.1 operates at 10 Gbps, great for flash storage. If you use Thunderbolt 3 (which uses a USB 3.1 type-C connector), then you get a PCIe 3.0 x4 connection which is around 32 Gbps, plus DisplayPort connectivity.

 

That said, the speed of the interface does not equate to the speed of the drive. If you plug in a slow drive through USB 3.1, it'll still be slow.

[patrickjp93]

4 hours ago, Bleedingyamato said:

What's an optane SSD?  

3DXPoint. About 3x faster than NVMe NAND.

[Bleedingyamato]

16 minutes ago, patrickjp93 said:

3DXPoint. About 3x faster than NVMe NAND.

From what I've been told even NVMe SSDs are so crazy fast that for most users they're so overkill as to be a semi-pointless waste of money.

 

They're also still much more expensive and smaller than standard SSDs if I'm not mistaken.

 

So while it's great there's a faster type of SSD storage is there a point to having something so fast that most people won't be able to afford or care about because it's too fast for their needs?

[MageTank]

1 hour ago, Bleedingyamato said:

From what I've been told even NVMe SSDs are so crazy fast that for most users they're so overkill as to be a semi-pointless waste of money.

 

They're also still much more expensive and smaller than standard SSDs if I'm not mistaken.

 

So while it's great there's a faster type of SSD storage is there a point to having something so fast that most people won't be able to afford or care about because it's too fast for their needs?

Niche markets are still markets nonetheless. The vast majority don't need them, but those that do will flock to the hardware that supports it. If it makes your product stand out in any way compared to the competition, its worth cornering that market. 

 

Most consumers don't do transfers that benefit from very high seq write speeds of NVMe drives, but the ones that do are grateful to have it. Then there are those that buy things they do not need/use, just for the bragging rights alone. I know plenty of people that own NVMe M.2 drives in ITX machines, and have 0 other drives in the machine. Making the entire point of having one completely useless. 

 

 

[Bleedingyamato]

10 minutes ago, MageTank said:

Niche markets are still markets nonetheless. The vast majority don't need them, but those that do will flock to the hardware that supports it. If it makes your product stand out in any way compared to the competition, its worth cornering that market. 

 

Most consumers don't do transfers that benefit from very high seq write speeds of NVMe drives, but the ones that do are grateful to have it. Then there are those that buy things they do not need/use, just for the bragging rights alone. I know plenty of people that own NVMe M.2 drives in ITX machines, and have 0 other drives in the machine. Making the entire point of having one completely useless. 

 

 

I just mean that a niche market doesn't exactly make for an easily profitable item.

 

If it costs a lot to make and few people buy it that seems very risky to me that the company making a product like that could set itself up to lose a lot of money.

 

Why does having only a NVMe drive defeat the point of having one?

[MageTank]

Just now, Bleedingyamato said:

I just mean that a niche market doesn't exactly make for an easily profitable item.

 

If it costs a lot to make and few people buy it that seems very risky to me that the company making a product like that could set itself up to lose a lot of money.

 

Why does having only a NVMe drive defeat the point of having one?

It's biggest boon is the fast sequential read and writes. Having only one drive means you are not writing to another, or reading from another drive. As far as overall responsiveness within the OS itself, it feels no different than any other SSD on the market either. 

 

As for the niche market and cost of implementing niche features, it's hard for us consumers to understand "why" it's done. Intel certainly has the funds to do it, but we do not know what goes on behind the scenes. Storage manufacturers could be making deals with Intel or motherboard manufacturers, paying them to add these features so they can potentially sell more drives, or it could just be added for what I've said before: to make your product look better by having more features. Intel has a pretty large stake in the storage solution world as they do make their own SSD's. Their SSD's are also considered some of the best, as no current prosumer SSD that I am aware of, can rival the 750 at queue depths. 

 

For businesses, the end must always justify the means. Why they are choosing to add it on a consumer platform is beyond me, but my understanding on the subject is quite limited anyways. 

[TetraSky]

So it's just the usual ~10% performances improvement we've been getting for the past several years eh...

Here's hoping AMD really shake things up on that side of things with Zen and the follow-up to Zen.

[NumLock21]

15 hours ago, Bleedingyamato said:

Question: Are the 7th generation i series CPUs going to be better enough versus the 6th to bother with if I already have a 6700K? (stock speed)

New products has to be better than the ones before. However, those improvements or new features it has, depends on whether you, see the needs to have it or not.

 

 

[Kenoxabe]

Zen, save us!

[patrickjp93]

53 minutes ago, TetraSky said:

So it's just the usual ~10% performances improvement we've been getting for the past several years eh...

Here's hoping AMD really shake things up on that side of things with Zen and the follow-up to Zen.

You try making old instructions execute in less than 1 cycle, because that's basically what Intel would have to do to get significant gains in old software. Everything else is hiding latency of memory and cache access at this point. This is why memory latency must come down AND bandwidth must come up. That's why the big moves in semiconductors right now are being made in memory and storage. Until you can better feed your logic circuitry, the gains made by latency hiding will be paltry, and re-optimizing with a better memory structure results in higher throughput at the same latency.

[Kenoxabe]

This is a serious question...

 

So are you guys gonna upgrade from Haswell Refresh to Kaby Lake considering the results?

[MageTank]

1 hour ago, TetraSky said:

So it's just the usual ~10% performances improvement we've been getting for the past several years eh...

Here's hoping AMD really shake things up on that side of things with Zen and the follow-up to Zen.

It's not even the usual 10% performance improvement. It's just improved efficiency, fin design and native USB 3.1/Thunderbolt/Optane. That's it. All of the additional performance Kaby has over Skylake, is the increased clock speed. IPC remains unchanged. 

 

32 minutes ago, patrickjp93 said:

You try making old instructions execute in less than 1 cycle, because that's basically what Intel would have to do to get significant gains in old software. Everything else is hiding latency of memory and cache access at this point. This is why memory latency must come down AND bandwidth must come up. That's why the big moves in semiconductors right now are being made in memory and storage. Until you can better feed your logic circuitry, the gains made by latency hiding will be paltry, and re-optimizing with a better memory structure results in higher throughput at the same latency.

That is my mantra, lol. 

 

[patrickjp93]

39 minutes ago, MageTank said:

It's not even the usual 10% performance improvement. It's just improved efficiency, fin design and native USB 3.1/Thunderbolt/Optane. That's it. All of the additional performance Kaby has over Skylake, is the increased clock speed. IPC remains unchanged. 

 

That is my mantra, lol. 

 

Not true. The updated Geekbench results show a 4% IPC increase in addition to the clock jump.

[Dabombinable]

On 03/10/2016 at 3:53 PM, MageTank said:

On Z board with "Multi Core Enhancement" (No idea what non-Asrock boards call it) enabled, they do. However, on my H170 board, my clocks fall back down to 4ghz under multi-core load. http://www.intel.com/content/www/us/en/support/processors/000005523.html

That's actually very weird. On my H87M Pro I was able to set MCE in the bios or within Windows, so that it would stay at 4.4GHz with all cores under a full load. This Z97 Sabertooth however, can't do that unless I install the version of Aisuite that came with the H87 board, or manually adjust the settings (aka, do an "overclock").

[MageTank]

5 minutes ago, Dabombinable said:

That's actually very weird. On my H87M Pro I was able to set MCE in the bios or within Windows, so that it would stay at 4.4GHz with all cores under a full load. This Z97 Sabertooth however, can't do that unless I install the version of Aisuite that came with the H87 board, or manually adjust the settings (aka, do an "overclock").

Odd. I need to look into it. My H170 Pro4 does not have the option for MCE at all. None of the "Advanced Turbo Features". Wonder if I can do some trickery with AMIflash on these H boards to see if its a simple flag to turn on?

[MageTank]

27 minutes ago, patrickjp93 said:

Not true. The updated Geekbench results show a 4% IPC increase in addition to the clock jump.

Where do you see that? I saw 3 different results last night, all drastically lower than the original 7700k result. Not only that, do we trust Geekbench's results when they are this inconsistent? I've proven myself that they are subject to change depending on multiple factors, including memory speed and amount of background tasks running. 

[Dabombinable]

29 minutes ago, patrickjp93 said:

Not true. The updated Geekbench results show a 4% IPC increase in addition to the clock jump.

Geekbench is as bad as WCCFTECH BTW....

[patrickjp93]

7 minutes ago, MageTank said:

Where do you see that? I saw 3 different results last night, all drastically lower than the original 7700k result. Not only that, do we trust Geekbench's results when they are this inconsistent? I've proven myself that they are subject to change depending on multiple factors, including memory speed and amount of background tasks running. 

That's up to the benchmarkers to handle correctly. There are two ddr4 2400 dual channel results I saw someone compare, and there's still a 3.8 (4)% IPC uplift.

 

6 minutes ago, Dabombinable said:

Geekbench is as bad as WCCFTECH BTW....

It's bad for comparing across ISAs, not within them. No benchmark is perfect.

[MageTank]

3 minutes ago, patrickjp93 said:

That's up to the benchmarkers to handle correctly. There are two ddr4 2400 dual channel results I saw someone compare, and there's still a 3.8 (4)% IPC uplift.

 

It's bad for comparing across ISAs, not within them. No benchmark is perfect.

"I saw someone compare" is not evidence. Give me more than that. Not only that, how did you know the memory speed? Do not even try to tell me you "extrapolated it from the bandwidth results" because the results are extremely inaccurate. My bandwidth is 30% slower on this test than any other bandwidth test I've used. 

 

There are only a few Geekbench submissions to choose from. In fact, I count exactly 8. https://browser.primatelabs.com/v4/cpu/search?utf8=✓&q=7700k

 

From these 8, only the Gigabyte results show a drastic improvement over Skylake. Feel free to pick the 4% improved IPC result out of those 8. Trust me, it doesn't exist because IPC was not changed.