Cheap Coffee Lake: Three Intel H370 and B360 Motherboards, Tested

Benchmark Results & Final Analysis

Synthetic Benchmarks

The point of including a Z370 motherboard in an H370/B360 roundup was to show how chipset selection should be based on features, not on some slim gradations of performance. The only performance benefits of the Z-series should be in overclocking and in high-data-rate memory support.

Strangely, the H370 boards outperformed both the B360 and Z370 in 3DMark. We should probably look at some other tests before drawing any conclusions.

The PCMark data we got from testing fits within the benchmark’s normal variations, Sandra’s CPU and DRAM tests show nothing askew, and everything’s normal in GPU-heavy Compubench. Let’s see what games tell us...

Game Testing

ASRock’s H370M-ITX/ac fell slightly behind at the lower settings of two benchmarks, but the other three boards battled it out to par. So the oddity in 3DMark Graphics scoring simply hasn’t shown up in real-world tests.

Timed Applications

Less completion time means more performance in timed benchmarks, and all of these show no noticeable performance variation based on chipset selection.

Power, Heat & Efficiency

We sometimes re-use data tables, and the ones used today were a copy of our Z370 data set. A funny thing happened when replacing the former system’s Core i7-8700K/GTX 1080 data with that of today’s Core i3-8350K/GTX 1050 Ti: The efficiency scores didn’t change. Depending on the motherboard, efficiency scores for the “lighter” test system were split around the average of the “heavy” system. After noticing that, we cleared the old data to get a clean look at the current data.

The Z system drew the least power at full CPU load, but that might be because the higher-priced boards typically use higher-quality voltage-regulation components. Looking at the H370 and B360 boards, the numbers appear to indicate that the smaller the board, the less energy it uses.

We should probably let you know now that these motherboards were not tested in the alphabetically organized order shown, nor in the reverse order, and appeared random when organized by form factor. The ASRock H370M-ITX/ac set our memory to 1.35V by default, and its added DIMM voltage probably has something to do with its higher CPU temperature.

Since all the boards in today’s comparison produced performance within 1% of each other, the efficiency chart is mostly a reflection of the average of full-load and no-load energy data. The H370 Aorus Gaming 3 WiFi has the most features of the B360 and H370 boards tested, and it also has the highest full-load power consumption.

Other Observations

As the chipset would indicate, only the Z-series board was overclock-capable, and base clock manipulation was the only way to overclock the RAM past DDR4-2400 on the tested Core i3-8350K. The CPU did surprisingly well, pushing our DDR4-3200 to DDR4-3504 despite the CPU's ostensible limits.

Since all of today’s boards performed similarly, the performance-per-dollar “value” chart simply reflects motherboard price. For around $100, the B360M Mortar has a little too much board for the chipset. (It could have used a couple more lanes.) Still, it seems a reasonable value in a MicroATX motherboard, if overclocking is not on your wish list, and you won't be tapping out all the lanes with a full complement of hardware.

In contrast to the MSI Mortar, for around $110, the ASRock H370M-ITX/ac has a little too much chipset for the board, leaving PCIe pathways unused. Dual Gigabit Ethernet and low-cost Wi-Fi appear to be worth more than its around-$10 price difference. But potential Z-series buyers hoping to save money by opting for an H-series board might be a little disappointed about its inclusion of only a single M.2 storage interface.

Meanwhile, the H370 Aorus Gaming 3 WiFi offers Intel’s 1.73Gb/s wireless controller and a bunch of RGB lighting for around $140. We'd willingly pay up to $130 to fill a specific need, and we wouldn’t be surprised if launch-week discounts (or subsequent ones) get this board to that price by the time you read this. At that price, for enthusiasts who want a keen combination of bling and connectivity features but not overclocking, this board looks to be a solid option.

Bear in mind, of course, that these are the first three new-chipset Coffee Lake boards that we got in hand, and we haven't tested any H310-based models yet at all. We expect to see further solid options in this arena as time goes on. So take these three boards in the early going simply as reasonable, on-point budget Coffee board picks--but far from the only ones that will emerge.

MORE: Best Motherboards

MORE: How To Choose A Motherboard

MORE: All Motherboard Content

This thread is closed for comments
23 comments
    Your comment
  • ghettogamer
    no under $80 mobo yet?
  • Crashman
    2430052 said:
    no under $80 mobo yet?
    We can see that MSI tried to push the top end of the B360 range with the Micro ATX sample submitted, so I wouldn't be surprised to see lesser B360 boards for $80. Below that will be H310, and I doubt anyone wants to show those off.
  • g-unit1111
    Now if we could just get cheap graphics cards we'd be back in business!
  • Saga Lout
    When the miners are staring at bankruptcy, they'll be glad to get a reasonable price.
  • 1_rick
    8708 said:
    2430052 said:
    no under $80 mobo yet?
    We can see that MSI tried to push the top end of the B360 range with the Micro ATX sample submitted, so I wouldn't be surprised to see lesser B360 boards for $80. Below that will be H310, and I doubt anyone wants to show those off.


    Newegg's got about a dozen sub-$80 boards, mostly H310, but several B360s.
  • Crashman
    1601753 said:
    8708 said:
    2430052 said:
    no under $80 mobo yet?
    We can see that MSI tried to push the top end of the B360 range with the Micro ATX sample submitted, so I wouldn't be surprised to see lesser B360 boards for $80. Below that will be H310, and I doubt anyone wants to show those off.
    Newegg's got about a dozen sub-$80 boards, mostly H310, but several B360s.
    Thanks Rick, I figured the bottom of the B360 market would be around $80 and with in-store discounts maybe a bit less. I don't think I'd recommend the H310 for much.
  • bit_user
    537231 said:
    Now if we could just get cheap graphics cards we'd be back in business!

    When is the last time you checked?

    Newegg has new GTX 1080's starting at $610, GTX 1060's from $270, and RX 580's from $350.

    Not exactly cheap, but prices have been dropping every week, for nearly a month.
  • Lutfij
    537231 said:
    Now if we could just get cheap graphics cards we'd be back in business!

    Amen to that brudda!

    When the miners are staring at bankruptcy, they'll be glad to get a reasonable price.
    Yeah, miners! :'(

    Nice article/review/writeup Thomas. Keep up the great work!
  • sunsanvil
    I would have been very interested in seeing you evaluate these boards with an upper end non-K processor. Its one thing to say that less expensive boards will tend to be coupled with lower end CPUs, but in the case of the Intel x70 family, the Z370 is all but pointless for anything which does not have a K at the end of it. That means that finding out how H and B boards make out with something like an i5-8600 or i7-8700 would be quite relevant, particularly as it relates to power delivery during all-core max turbo situations.
  • Crashman
    344126 said:
    I would have been very interested in seeing you evaluate these boards with an upper end non-K processor. Its one thing to say that less expensive boards will tend to be coupled with lower end CPUs, but in the case of the Intel x70 family, the Z370 is all but pointless for anything which does not have a K at the end of it. That means that finding out how H and B boards make out with something like an i5-8600 or i7-8700 would be quite relevant, particularly as it relates to power delivery during all-core max turbo situations.
    There are of course other reasons to use a Z, including support for DDR4-3200 etc, SLI/Crossfire, and CPU-based NVMe.

    Your point is also appropriate, and the combination of these two points could leave one question why I didn't use the 8700K in the first place and tell readers "if the board can handle this it can certainly handle that". We used to do things that way until angry budget-gamers came in and said "we don't use expensive CPUs on cheap motherboards". And so we'd be left testing the boards on at least two processors...and heck the higher one might as well be the 8700K since we already have it right? And we'd at least get to show whether or not these boards can lock-in the highest "stock" Turbo Boost ratio with that one, right?
  • sunsanvil
    8708 said:
    (...) and tell readers "if the board can handle this it can certainly handle that"


    Personally that's what I would like to see, but I understand the arduous task of dealing with a global peanut gallery. :)

    Of the crop featured here the MSI is the most interesting to me in that, anecdotally at least, I see an awful lot of builds go their entire life with nothing beyond CPU, RAM, a single GPU, a SATA drive or two, and more recently an M.2 maybe, yet people still want the highest build quality and core performance.
  • gaurav71189
    A correction needed:

    "the remaining six HSIO serve the three USB 3.1 Gen2 ports"

    There are only 2 USB 3.1 Gen2 ports (one type A and one type C at the back). Only one HSIO is used for the third USB 3.1 Gen1 port that is available as a header for type C. I believe the remaining 1 HSIO lane would be allocated to the GbE port. I'm not sure if the CNVi uses a lane.
  • Crashman
    1857486 said:
    A correction needed: "the remaining six HSIO serve the three USB 3.1 Gen2 ports" There are only 2 USB 3.1 Gen2 ports (one type A and one type C at the back). Only one HSIO is used for the third USB 3.1 Gen1 port that is available as a header for type C. I believe the remaining 1 HSIO lane would be allocated to the GbE port. I'm not sure if the CNVi uses a lane.
    Well what do you know, I must have temporarily mistaken it for an honest USB 3.1 Gen2 front-panel port rather than a sham version. As for the Ethernet, it's an i219V so it has an independent path and...oh crap, I checked the original table against the version you're seeing, they REMOVED i219V and replaced it with the word "Intel", as if that was useful. Thanks for asking a question I'd already answered, otherwise I wouldn't have looked for the error.

    See if this makes more sense ;)
    "The first PCIe x16 slot is fed by the CPU’s PCIe controller and doesn’t count toward the chipset’s 30-lane limit. The second x16-length slot has four chipset lanes, the four x1 slots have fixed pathways, and only one of the six SATA ports is a potential share (in the unlikely event that it gets lost to a SATA-based M.2 card). The upper and lower M.2 drive slots consume four and two lanes (respectively), the two rear-panel USB 3.0 ports takes up two more HSIO resources, the front-panel USB 3.0 header consumes only one HSIO through a USB hub, the USB 3.1 front-panel header consumes one HSIO since it's only connected to a Gen1 interface, and our math says that the four of the remaining six HSIO serve rear-panel USB 3.1 Gen2 ports. The two remaining HSIO pathways would seam to go to an unused M.2 Key-E interface, since the CNVi connector is cross-compatible with Key-E devices. All that said, for a budget-minded PC, the second (two-path) M.2 storage slot may likely go unused, rendering its lane count moot. "
  • gaurav71189
    8708 said:
    1857486 said:
    A correction needed: "the remaining six HSIO serve the three USB 3.1 Gen2 ports" There are only 2 USB 3.1 Gen2 ports (one type A and one type C at the back). Only one HSIO is used for the third USB 3.1 Gen1 port that is available as a header for type C. I believe the remaining 1 HSIO lane would be allocated to the GbE port. I'm not sure if the CNVi uses a lane.
    Well what do you know, I must have temporarily mistaken it for an honest USB 3.1 Gen2 front-panel port rather than a sham version. As for the Ethernet, it's an i219V so it has an independent path and...oh crap, I checked the original table against the version you're seeing, they REMOVED i219V and replaced it with the word "Intel", as if that was useful. Thanks for asking a question I'd already answered, otherwise I wouldn't have looked for the error. See if this makes more sense ;) "The first PCIe x16 slot is fed by the CPU’s PCIe controller and doesn’t count toward the chipset’s 30-lane limit. The second x16-length slot has four chipset lanes, the four x1 slots have fixed pathways, and only one of the six SATA ports is a potential share (in the unlikely event that it gets lost to a SATA-based M.2 card). The upper and lower M.2 drive slots consume four and two lanes (respectively), the two rear-panel USB 3.0 ports takes up two more HSIO resources, the front-panel USB 3.0 header consumes only one HSIO through a USB hub, the USB 3.1 front-panel header consumes one HSIO since it's only connected to a Gen1 interface, and our math says that the four of the remaining six HSIO serve rear-panel USB 3.1 Gen2 ports. The two remaining HSIO pathways would seam to go to an unused M.2 Key-E interface, since the CNVi connector is cross-compatible with Key-E devices. All that said, for a budget-minded PC, the second (two-path) M.2 storage slot may likely go unused, rendering its lane count moot. "


    There seems to be a confusion again with the H370!

    What I meant to say was:
    2 lanes for 2 USB 3.1 Gen1 ports at the back
    2 lanes for 2 USB 3.1 Gen1 ports header on board
    1 lane for 1 USB 3.1 Gen1 Type-C header on board
    4 lanes for 2 USB 3.1 Gen2 ports at the back (one type-A and one type-C)
    And the remaining PCIe (4 + 1 + 1 + 1 + 1) + SATA (6) + M.2 (4 + 2)

    This brings the count to 29 lanes. That would leave 1 port unaccounted for, unless it's used for either the GbE or the CNVi. Or it may even be shared? I'm not sure which one would account for the remaining 1.

    I hope you've now understood my concern. I bought the H370 Aorus WiFi just a few days ago. Wonderful board though.
  • Crashman
    1857486 said:
    8708 said:
    1857486 said:
    A correction needed: "the remaining six HSIO serve the three USB 3.1 Gen2 ports" There are only 2 USB 3.1 Gen2 ports (one type A and one type C at the back). Only one HSIO is used for the third USB 3.1 Gen1 port that is available as a header for type C. I believe the remaining 1 HSIO lane would be allocated to the GbE port. I'm not sure if the CNVi uses a lane.
    Well what do you know, I must have temporarily mistaken it for an honest USB 3.1 Gen2 front-panel port rather than a sham version. As for the Ethernet, it's an i219V so it has an independent path and...oh crap, I checked the original table against the version you're seeing, they REMOVED i219V and replaced it with the word "Intel", as if that was useful. Thanks for asking a question I'd already answered, otherwise I wouldn't have looked for the error. See if this makes more sense ;) "The first PCIe x16 slot is fed by the CPU’s PCIe controller and doesn’t count toward the chipset’s 30-lane limit. The second x16-length slot has four chipset lanes, the four x1 slots have fixed pathways, and only one of the six SATA ports is a potential share (in the unlikely event that it gets lost to a SATA-based M.2 card). The upper and lower M.2 drive slots consume four and two lanes (respectively), the two rear-panel USB 3.0 ports takes up two more HSIO resources, the front-panel USB 3.0 header consumes only one HSIO through a USB hub, the USB 3.1 front-panel header consumes one HSIO since it's only connected to a Gen1 interface, and our math says that the four of the remaining six HSIO serve rear-panel USB 3.1 Gen2 ports. The two remaining HSIO pathways would seam to go to an unused M.2 Key-E interface, since the CNVi connector is cross-compatible with Key-E devices. All that said, for a budget-minded PC, the second (two-path) M.2 storage slot may likely go unused, rendering its lane count moot. "
    There seems to be a confusion again with the H370! What I meant to say was: 2 lanes for 2 USB 3.1 Gen1 ports at the back 2 lanes for 2 USB 3.1 Gen1 ports header on board 1 lane for 1 USB 3.1 Gen1 Type-C header on board 4 lanes for 2 USB 3.1 Gen2 ports at the back (one type-A and one type-C) And the remaining PCIe (4 + 1 + 1 + 1 + 1) + SATA (6) + M.2 (4 + 2) This brings the count to 29 lanes. That would leave 1 port unaccounted for, unless it's used for either the GbE or the CNVi. Or it may even be shared? I'm not sure which one would account for the remaining 1. I hope you've now understood my concern. I bought the H370 Aorus WiFi just a few days ago. Wonderful board though.

    Please re-read what I quoted from the updated article:
    4 lanes for the third x16 slot
    4 lanes for the three x1 slots
    6 lanes for SATA
    4 lanes for the main M.2 storage slot
    2 lanes for the second M.2 storage slot
    2 lanes for the rear USB 3.0
    1 lane for the front (dual port) USB 3.0 because it's on a hub
    1 lane for the front (single port) USB 3.1 because its Gen1
    4 lanes for the two REAR USB 3.1 Gen2 ports (Type A, C)
    2 lanes for M.2 Key-E Wi-Fi card compatibility, even though they're not used.

    4+4+6+4+2+2+1+1+4+2=30

    The CNVi module doesn't require HSIO, but the slot itself is M.2 Key-E compatible so it still needs two PCIe pathways.

    I hope I addressed your concern two posts ago ;)
  • Crashman
    1857486 said:
    I hope you've now understood my concern. I bought the H370 Aorus WiFi just a few days ago. Wonderful board though.
    What I'm saying is that your comments prompted me to recalculate the entire thing using missed information, and even though we didn't come up with the same answer, I'm crediting you, right here, for prompting me to do that recalculation.
  • bit_user
    You seem to be on quite the charm offensive.
    8708 said:
    Thanks for asking a question I'd already answered, otherwise I wouldn't have looked for the error.

    I'm simply dazzled.

    8708 said:
    I hope I addressed your concern two posts ago ;)

    Being right doesn't negate the benefits of being nice.
  • Crashman
    328798 said:
    You seem to be on quite the charm offensive.
    8708 said:
    Thanks for asking a question I'd already answered, otherwise I wouldn't have looked for the error.
    I'm simply dazzled.
    8708 said:
    I hope I addressed your concern two posts ago ;)
    Being right doesn't negate the benefits of being nice.

    Doh! But if he hadn't mentioned the codec I wouldn't have found the info missing in the table! And he appeared not to read the new quote in his next response, prompting me to explain the quote in further detail. And his questions did help me to improve the article. And I even thanked him, so maybe I'm just not very good at being nice?
  • bit_user
    8708 said:
    maybe I'm just not very good at being nice?

    I'm making rather too big a deal out of those couple lines. If I were unfamiliar with your acerbic tone, I might not have picked up on them.

    I apologize for overreacting. You took that rather better than I expected.
  • gaurav71189
    8708 said:
    Please re-read what I quoted from the updated article: 4 lanes for the third x16 slot 4 lanes for the three x1 slots 6 lanes for SATA 4 lanes for the main M.2 storage slot 2 lanes for the second M.2 storage slot 2 lanes for the rear USB 3.0 1 lane for the front (dual port) USB 3.0 because it's on a hub 1 lane for the front (single port) USB 3.1 because its Gen1 4 lanes for the two REAR USB 3.1 Gen2 ports (Type A, C) 2 lanes for M.2 Key-E Wi-Fi card compatibility, even though they're not used. 4+4+6+4+2+2+1+1+4+2=30 The CNVi module doesn't require HSIO, but the slot itself is M.2 Key-E compatible so it still needs two PCIe pathways. I hope I addressed your concern two posts ago ;)


    Thanks, I've now understood correctly. I'd misunderstood some of the points and was under the impression that the USB 3.1 Gen1 Header was not in a hub and that it used 2 full lanes. Also that the CNVi port (M.2 Key-E), as you mentioned has to use 2 lanes is what I've understood after your detailed explanation.

    8708 said:
    What I'm saying is that your comments prompted me to recalculate the entire thing using missed information, and even though we didn't come up with the same answer, I'm crediting you, right here, for prompting me to do that recalculation.


    As long as it leads to the correctness of the calculations, we're both happy :)
  • Crashman
    1857486 said:
    8708 said:
    Please re-read what I quoted from the updated article: 4 lanes for the third x16 slot 4 lanes for the three x1 slots 6 lanes for SATA 4 lanes for the main M.2 storage slot 2 lanes for the second M.2 storage slot 2 lanes for the rear USB 3.0 1 lane for the front (dual port) USB 3.0 because it's on a hub 1 lane for the front (single port) USB 3.1 because its Gen1 4 lanes for the two REAR USB 3.1 Gen2 ports (Type A, C) 2 lanes for M.2 Key-E Wi-Fi card compatibility, even though they're not used. 4+4+6+4+2+2+1+1+4+2=30 The CNVi module doesn't require HSIO, but the slot itself is M.2 Key-E compatible so it still needs two PCIe pathways. I hope I addressed your concern two posts ago ;)
    Thanks, I've now understood correctly. I'd misunderstood some of the points and was under the impression that the USB 3.1 Gen1 Header was not in a hub and that it used 2 full lanes. Also that the CNVi port (M.2 Key-E), as you mentioned has to use 2 lanes is what I've understood after your detailed explanation.
    8708 said:
    What I'm saying is that your comments prompted me to recalculate the entire thing using missed information, and even though we didn't come up with the same answer, I'm crediting you, right here, for prompting me to do that recalculation.
    As long as it leads to the correctness of the calculations, we're both happy :)

    Sorry for the misunderstanding:
    1.) CNVi doesn't use any lanes. CNVi uses a combination connector that has two interfaces, and the second interface is M.2 Key-E.

    2.) M.2 Key-E has two PCIe lanes for Wi-Fi and USB 2.0 for Bluetooth.

    3.) Since the board has combination slot compatible with both Key-E and CNVi devices, it has 1x CNVi, 2xPCIe, and 1xUSB 2.0 interfaces combined on that slot.

    4.) Since the board comes with a CNVi module, the two PCIe pathways that go to the slot aren't used by the module.

    5.) The unused PCIe pathways get used if you remove the CNVi module and replace it with an M.2 Key-E module.
  • gaurav71189
    8708 said:
    Sorry for the misunderstanding: 1.) CNVi doesn't use any lanes. CNVi uses a combination connector that has two interfaces, and the second interface is M.2 Key-E. 2.) M.2 Key-E has two PCIe lanes for Wi-Fi and USB 2.0 for Bluetooth. 3.) Since the board has combination slot compatible with both Key-E and CNVi devices, it has 1x CNVi, 2xPCIe, and 1xUSB 2.0 interfaces combined on that slot. 4.) Since the board comes with a CNVi module, the two PCIe pathways that go to the slot aren't used by the module. 5.) The unused PCIe pathways get used if you remove the CNVi module and replace it with an M.2 Key-E module.


    Quite a confusing set of stuff! Got it now, thanks again :) So, that means 2x PCIe is potentially unused.
  • Crashman
    1857486 said:
    8708 said:
    Sorry for the misunderstanding: 1.) CNVi doesn't use any lanes. CNVi uses a combination connector that has two interfaces, and the second interface is M.2 Key-E. 2.) M.2 Key-E has two PCIe lanes for Wi-Fi and USB 2.0 for Bluetooth. 3.) Since the board has combination slot compatible with both Key-E and CNVi devices, it has 1x CNVi, 2xPCIe, and 1xUSB 2.0 interfaces combined on that slot. 4.) Since the board comes with a CNVi module, the two PCIe pathways that go to the slot aren't used by the module. 5.) The unused PCIe pathways get used if you remove the CNVi module and replace it with an M.2 Key-E module.
    Quite a confusing set of stuff! Got it now, thanks again :) So, that means 2x PCIe is potentially unused.

    Right, it means its unused by default and available to a Key-E card if you ever feel the need to replace the CNVi with Key-E. Perhaps Gigabyte will have a cheaper model based on the same circuit board that uses the old 433Mbit Key-E module...or perhaps maintaining compatibility was a priority. The later would necessarily be true if Gigabyte sells a cheaper version of the board with an empty module slot.