it isn't mandatory. USB 3.1 is added via a third party controller. Nothing would stop a motherboard manufacturer from not adding it other than the prospect of such a board not selling.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Intel Skylake Core i7-6700K IPC & Overclocking Review @ [H]
- Thread starter FrgMstr
- Start date
it isn't mandatory. USB 3.1 is added via a third party controller. Nothing would stop a motherboard manufacturer from not adding it other than the prospect of such a board not selling.
TroubleMagnet
Gawd
- Joined
- Oct 19, 2005
- Messages
- 535
We've essentially hit a speed limit with Silicon around 5GHz. You can get past it with heroic cooling efforts, but only so much. Voltage scaling is also dead, it used to drop with each generation as well but we've hit a wall there as well. Some new materials (hybrid silicon, or different Gallium based mixes) have promise for at least getting clock speeds higher but geometric clock speed increases are unlikely to return.
Single thread performance has gone from an exponential increase to a linear one and has been there for a while now. Parallel performance does a bit better since we're still getting more transistors each generation but due to diminishing returns it's also linear, just a big steeper slope than single threaded.
Single thread performance has gone from an exponential increase to a linear one and has been there for a while now. Parallel performance does a bit better since we're still getting more transistors each generation but due to diminishing returns it's also linear, just a big steeper slope than single threaded.
![Zarathustra[H]](https://cdn.hardforum.com/data/avatars/m/9/9298.jpg?1707758471){kind=avatar}
Zarathustra[H]
Extremely [H]
- Joined
- Oct 29, 2000
- Messages
- 38,848
obiwansotti
Gawd
- Joined
- Dec 13, 2002
- Messages
- 1,003
Zarathustra[H];1041787873 said:This I consider to be a red herring.
What a USB type C connector? Just get an adapter.
Type C != 3.1
3.1 has 2x the bandwidth of 3.0,
You certainly can make a 3.0 port into a type-C connector, but the bandwidth and power capability won't improve.
Zarathustra[H]
Extremely [H]
- Joined
- Oct 29, 2000
- Messages
- 38,848
Ahh, fair enough. I didn't realize 3.1 was a major step. usually the point releases are not huge like that.
That being said, I don't really know what I use USB for that would benefit greatly from higher bandwidth or power, but I'm sure its important to someone.
Type-C also has an alternate mode which allows additional signals to pass through it. It can pass Thunderbolt 3.0 over it which has up to 40GB/s of bandwidth. VESA signals can also pass through it. It's actually a pretty universal connector that could theoretically replace most ports on our systems if it gains acceptance. Unlike previous USB standards it is also reversible. You no longer have to keep flipping the bastard until it goes in.
Zarathustra[H];1041788023 said:
The majority of USB C interfaces seem to be targeted towards mobile computing. Smaller format with higher power so they can have external devices connected through a single low profile connector.
I have very little use for USB on my desktop other than keyboard and mouse. I have an internal card reader for all the camera memory types I use and I occasionally use an external USB drive.
obiwansotti
Gawd
- Joined
- Dec 13, 2002
- Messages
- 1,003
I believe the alpine ridge chip that intel is going to ship for USB 3.1 will do displayport 1.3, thurderbolt 3.0, HDMI 2.0 w/ HDCP 2.2 pretty gawd damn universal.
That's the one you really want.
obiwansotti
Gawd
- Joined
- Dec 13, 2002
- Messages
- 1,003
I'm sure your info is better. I was just trying to show how flexible alpine ridge will be. Makes even the current USB 3.1 type-c functionality seem limited.
obiwansotti
Gawd
- Joined
- Dec 13, 2002
- Messages
- 1,003
I can't imagine that's a thing, it wouldn't be a type-C connector as that connector is physical reversible.
You put USB 2.0 on a type-c connector like mentioned above, it's just the physical connection for the wires.
You can do 10g Ethernet emulation over it as well. The connection is very versatile and if they don't alter the physical connection we will inevitably see enhancements to the specification over time.
My local MicroCenter in VA said they expected the i7-6700k at the end of the month.
Ohio store told me the same thing, but I have seen stuff show up sooner than they said.
I should not really be in too much of a hurry because some of the boards are not yet available and I have not decided which version I want. I will most likely stick to Asus, but they have so many versions of the board even within a specific brand it is kind of insane.
I have new everything sitting waiting on CPU, MB and RAM.
How important do you think having the actual Alpine Ridge chipset is to full USB 3.1 spec coverage? Will these third-party chipsets cover all the things like video and thunderbolt or should we be sticking to the official Intel chipset?
Just wondering as I'm weighing all options for a board that I hope to last for a very long time, and AFAIK only the Gigabyte boards actually have Alpine Ridge at present.
I've been leaning toward MSI but their own website's documentation is a mess and I've been relying on reviews. Staying away from ASUS.
Just wondering as I'm weighing all options for a board that I hope to last for a very long time, and AFAIK only the Gigabyte boards actually have Alpine Ridge at present.
I've been leaning toward MSI but their own website's documentation is a mess and I've been relying on reviews. Staying away from ASUS.
obiwansotti
Gawd
- Joined
- Dec 13, 2002
- Messages
- 1,003
None of the current chipsets do any of that stuff.
AFAIK intel is the only one to make any thunderbolt chips. If that trend continues, the thunderbolt, display port, HDMI output for type-C will require the an Alpine Ridge or it's successor.
Since I am also trying to figure out what motherboard is best I am interested in why you are staying away from ASUS? I got the impression that they really missed the mark on the x99 boards, but my experience with them has always been good.
I have not nailed down what specific features I feel like I need, but one of the things I do like about ASUS is the inclusion of numerous 4pin PWM headers for my fans. No this isnt critical there are many ways to control fans, but I happen to have 4 PWM fans in my case and ASUS seems to have a nice built in solution.
In general I like their BIOS and the auto settings for overclock
obiwansotti
Gawd
- Joined
- Dec 13, 2002
- Messages
- 1,003
And that's stock for stock.
If the 860 can do 4.2-4.3ish, then gap to an o/c'd 6700K shrinks to more like 50-60%
I think skylake is making us old school overclockers face facts that CPU performance has finally hit a wall. Instead of getting double the perfomance every 2 years, it's gonna be 6 or 7, or maybe longer.
For desktop use, does it matter? Normally, when I'm doing a CPU upgrade, there's some serious slowness in the system. With SSDs and massive amounts of ram, my system isn't slow. It's lacking some features, but CPU speed is rarely an issue. I grant you that many may use apps that need more processing power, but if I don't need it, then that makes it a small part of a small part of the buyers.
Sure I wish it was 10x faster, but I'm not sure I'd notice a difference. As it is simple desktop stuff launches in a second. Not bad for a system that's dangerously close to 6 year s old.
Honestly, if there were new MB's with updated features, I'd consider just buying a new MB (but I'd probably still buy a new CPU
). It's funny. We use to bitch about bloatware, now we bitch that CPUs aren't getting faster fast enough, even though 99.9% of the s/w out there has no need for a faster processor.
The only thing that I have that needs it is Lightroom and PS and with Lightroom, I'm convinced it's poor coding, because I've heard that my complaints exist on the latest hardware.
I have been on the same evga x58 board (rev 1.0, so I can't use a Xeon without a mod) for almost 7 years. But before that I had Asus boards with capacitor whine and multiple issues. Dealt with them in an RMA about 12 years ago that made me pull my hair out.
Would have considered them again but just reading about the problems with their warranty in the Motherboards forum makes me shy away.
Really liking the MSI boards in terms of look/function but if Gigabyte is the only one with Alpine Ridge in the near future I have to consider them.
They just have an awful look (IMO) and really bad type on the boards. Shouldn't matter since I have red LEDs in front of the board anyway and will rarely even look at it, but I'm just a bit gunshy with them.
Would have considered them again but just reading about the problems with their warranty in the Motherboards forum makes me shy away.
Really liking the MSI boards in terms of look/function but if Gigabyte is the only one with Alpine Ridge in the near future I have to consider them.
They just have an awful look (IMO) and really bad type on the boards. Shouldn't matter since I have red LEDs in front of the board anyway and will rarely even look at it, but I'm just a bit gunshy with them.
Keep in mind that having Alpine Ridge onboard guarantees nothing except USB 3.1 support. Motherboard manufacturers do not have to support UASP over USB 3.0, they do not have to support Thunderbolt 3.0, VESA signals / DisplayPort or HDMI over USB. These things are optional and are subject to certification from Intel and requires appropriate firmware to implement.
The ASMedia ASM1142 can handle UASP and USB 3.1. According to ASUS the ASM1142 is just as fast as Alpine Ridge. I haven't tested that yet. But basically if that's true or they are close it doesn't matter when comparing motherboards without Thunderbolt 3.0 certification. The ASM1142 will be cheaper. Though ASUS' motherboards aren't going to be necessarily. They have some extra hardware on them compared to other boards. Fan control ICs, more clock generators and the like. Features like their fan controls require additional hardware to implement.
In other words they aren't cheaper to build. ASUS just allocates their money differently. Gigabyte on the other hand is using Alpine Ridge on most if not all their boards this time. They are also using PLX chips on many models. I don't know how they intend to keep costs down.
They all cut corners. The trick is seeing who cuts corners in the places that you can live with.
Last edited:
Thanks for the info, I have an ASUS deluxe board my I7-920 has been running on for 6 years now without issue. MSI has looked strong lately so I might consider them as well.
I can live without the Thunderbolt, just trying to consider everything.
Ultimately I tend to lean toward the OC boards figuring they have better components to be stable for a long time. Always an educated crap shoot.
Been really interested in the MSI Z170 XPower but also looking at their M7.
Ultimately I tend to lean toward the OC boards figuring they have better components to be stable for a long time. Always an educated crap shoot.
Been really interested in the MSI Z170 XPower but also looking at their M7.
Aye.
Looks like this is really an extraordinarily compelling upgrade for people still running LGA775!
Commander Suzdal
2[H]4U
- Joined
- Jul 29, 2004
- Messages
- 2,134
Yeegads, yes! Perhaps even gadzooks! Theoretically there is at least a 50% chance of me getting the "flip" right on the first try, but my success rate is more like 1/5. And then there's the times that I got it right the first time, but flipped it anyway because I didn't hold it at just the right angle while trying to insert it. So I flip, then flip back. So tired of it.
I have been looking at the ASUS ROG MAXIMUS VIII HERO. The MSI M7 would be my 2nd choice if I didn't go with ASUS.
The XPOWER looks interesting, but I am still trying to figure out why it is more expensive than the M7. What is it giving me? The compare on the site isnt as clear as I would like. Th Asus compare doesnt even work for me.
From what information I have, it looks like the XPower Gaming Titanium has the onboard OC panel controls, an Intel NIC and a PLX chip. The PLX chip is the main thing that will drive the cost up.
Assuming I don't need the extra PCIe lanes which I wont since I am not going to use SLI, isn't it better to not have the PLX chip?
I thought I had read that the PLX chip can cause some added latency. I guess at a minimum its cost that I do not need.
Depends, those PCIe lanes can be used for SATA/USB/m.2 data, leaving your GPU x16 slots free.
PLX chips do in fact create increased latency. If you aren't going to use a multi-GPU solution, or even if you aren't going to go beyond two graphics cards a PLX chip is a waste of money.
The strange thing that the MSI Z170 Xpower is missing is the USB 3.1 type C, the small reversible connector. It's the only one of the higher-end boards missing it, for whatever reason.
Thing is, I don't know how often I'd ever really need to use the type C in back of my desktop ... when I want to plug anything in back there I have to crawl under my desk anyway so whether it's type A or type C seems kinda meaningless as long as they're both 3.1.
Never bought MSI before and the number of spelling mistakes and errors on their official website is kinda concerning. Hell, they have a vertically-oriented USB port on the back of their board mislabeled as HDMI. They also have a label that says USB 3.1 type C pointing to what is clearly a type A. Hopefully the images on the site are just placeholder and the actual board will differ when it ships ... which won't be until Sept.
Thing is, I don't know how often I'd ever really need to use the type C in back of my desktop ... when I want to plug anything in back there I have to crawl under my desk anyway so whether it's type A or type C seems kinda meaningless as long as they're both 3.1.
Never bought MSI before and the number of spelling mistakes and errors on their official website is kinda concerning. Hell, they have a vertically-oriented USB port on the back of their board mislabeled as HDMI. They also have a label that says USB 3.1 type C pointing to what is clearly a type A. Hopefully the images on the site are just placeholder and the actual board will differ when it ships ... which won't be until Sept.
I am with you on the new USB connector. Part of me thinks its a good idea to have newer tech just in case something cool comes of it, but odds are I will never use it. I really think the point of it is to have an all in one type connector for laptops that saves space.
I am going to steer clear of MB with PLX chips, seems a waste of money for my needs. With a single video card I should still have enough lanes if I wanted to move to a PCIe boot SSD when prices come down. I considered it, but when I can get 500gb for $150 vs 400gb for $380 and I am not sure the faster drive would have enough impact for what I use the system for.
I have only used MSI for video cards, but they seem to have good reviews for their MBs. Lets face it we have seen bad boards from all kinds of manufacturers so its always a little bit of a guessing game. I know someone who loves and hates just about every brand.
Good luck, now someone make some I7s appear!
Aye.
Looks like this is really an extraordinarily compelling upgrade for people still running LGA775!
As somebody finally upgrading from LGA775 yes, yes it is
MTDEW
Weaksauce
- Joined
- Oct 30, 2013
- Messages
- 123
After reading this review, I decided to test my own 2600k system in Metro LL and see how my results compared to the ones in the review since I'm running 16gb of DDR3 1600 vs the 16gb DDR3 1866 used for this review on their 2600k.
Now my Win7 Ultimate 64bit score was what I expected for DDR3 1600 vs DDR3 1866 , but what I didn't expect was the much poorer score using Windows 10 Pro 64bit!
My test setup.
Win7 Ultimate 64bit / Win10 Pro 64bit
video driver: 15.7.1 on both OS's
i7 [email protected] (Noctua NH-D14)
16gb G-skill DDR3 1600 (24-9-9-9)
Asus P8Z68-V Pro mobo
Sapphire r9 290 Tri-x
Samsung 840 Pro 250gb
WD Blue 1tb / WD Black 640gb
Asus Xonar DG soundcard
Corsair HX850 PSU
Note: I use Acronis True Image to make an image of my OS's when I do a clean install, so these tests were done using a clean install of both OS's
Metro LL Benchmark (default benchmark with sound disabled via Device manager)
1024x768 (low settings)
Win7 Ultimate 64bit - avg 157fps
Win10 Pro 64 bit - avg 139fps
Now this is on the same system with same hardware.
Up until now I've been very happy with my Win10 gaming performance and still am actually.
But the above numbers do bother me a bit.
Can anyone else confirm the same CPU performance difference using Win7 vs Win10 when testing with the Metro LL Benchmark?
Now my Win7 Ultimate 64bit score was what I expected for DDR3 1600 vs DDR3 1866 , but what I didn't expect was the much poorer score using Windows 10 Pro 64bit!
My test setup.
Win7 Ultimate 64bit / Win10 Pro 64bit
video driver: 15.7.1 on both OS's
i7 [email protected] (Noctua NH-D14)
16gb G-skill DDR3 1600 (24-9-9-9)
Asus P8Z68-V Pro mobo
Sapphire r9 290 Tri-x
Samsung 840 Pro 250gb
WD Blue 1tb / WD Black 640gb
Asus Xonar DG soundcard
Corsair HX850 PSU
Note: I use Acronis True Image to make an image of my OS's when I do a clean install, so these tests were done using a clean install of both OS's
Metro LL Benchmark (default benchmark with sound disabled via Device manager)
1024x768 (low settings)
Win7 Ultimate 64bit - avg 157fps
Win10 Pro 64 bit - avg 139fps
Now this is on the same system with same hardware.
Up until now I've been very happy with my Win10 gaming performance and still am actually.
But the above numbers do bother me a bit.
Can anyone else confirm the same CPU performance difference using Win7 vs Win10 when testing with the Metro LL Benchmark?
Kyle_Bennet - nice article! very informative.
Seems super important that we pay attention to RAM speed when purchasing as noted in your article.
What about RAM timings? unless i'm mistaken, they are not covered in the article.
I'm assuming that raw bandwidth is more important than tight timings?
Thanks.
Seems super important that we pay attention to RAM speed when purchasing as noted in your article.
What about RAM timings? unless i'm mistaken, they are not covered in the article.
I'm assuming that raw bandwidth is more important than tight timings?
Thanks.