mobomaps.com

Compare motherboards by tracing where each PCIe lane comes from (CPU or chipset) and where it goes (which slot), then check lane sharing before you buy. Spec counts alone won't tell you whether populating a second M.2 slot drops your GPU from x16 to x8 — a lane map will. This matters most if you plan to fill more than one M.2 slot, add a capture or network card, or use USB4 alongside a full-speed GPU.

Why spec sheets aren't enough

Two boards can both list "5 M.2 slots" and "2 PCIe x16 slots" and behave completely differently once you install devices. The difference is routing:

• CPU-direct lanes are the fastest and least contended. On a typical AM5 or LGA1851 platform, the CPU provides roughly 20–24 usable PCIe lanes. The usual split is x16 for the primary GPU slot and x4 for the first NVMe SSD.
• Chipset lanes expand capacity but share a narrower uplink back to the CPU. Devices on chipset lanes can contend with each other and with chipset-attached USB.

A board with more slots isn't automatically better — it may just be splitting the same lanes more ways.

Step 1: Trace every lane from source to slot

For each board you're considering, map out:

The key question for each slot: does populating it change the width of another slot?

Step 2: Check lane sharing and bandwidth conflicts

Lane sharing is when two or more slots use the same physical lanes. A concrete example from a mapped X870E board: an M.2 slot shares lanes with USB4, so enabling both splits bandwidth between them. On other boards, installing a drive in a particular M.2 slot drops the GPU slot from x16 to x8.

Before buying, list your planned devices and toggle them against the board's lane map:

1. GPU in the primary x16 slot
2. Primary NVMe in the CPU-direct M.2
3. Each additional NVMe, capture card, or network card
4. USB4 devices, if you rely on them

If any combination forces a slot you need down to x8 or x2, that's the conflict you'd otherwise discover after building.

Step 3: Match the lane layout to your actual device plan

Work backward from what you'll install, not from the maximum the board supports:

• GPU + one NVMe only: almost any modern board works. You need 20 lanes minimum (x16 + x4).
• GPU + two or three NVMe drives: check whether the second and third M.2 slots are CPU-direct, chipset-attached, or shared. Shared slots may run at x2.
• GPU + NVMe + capture or 10G network card: confirm the secondary PCIe slot is electrically x4 or better and doesn't share with an M.2 you're using.
• USB4 plus multiple drives: verify no M.2 slot shares lanes with the USB4 controller.

Step 4: Weigh the lane budget against chipset tier

CPUs provide 20–24 usable lanes; the chipset expands from there. Higher-tier chipsets (X870E, Z890) generally expose more chipset lanes and more USB4, but the CPU-direct allocation is similar across a generation. So the practical comparison is:

• How many CPU-direct slots does the board give you, and at what width?
• How many chipset lanes, and do they contend with each other?

A cheaper board with clean routing can beat a premium board with heavy sharing for a specific build.

Step 5: Compare boards side by side on the specs that matter

When stacking two boards, compare these dimensions directly rather than the full spec list:

For example, comparing an X870E board with 5 M.2 slots and 2 USB4 ports against one with 4 M.2 slots and 2 USB4 ports: the extra M.2 may come at the cost of a shared lane, so the count alone doesn't settle it.

Common pitfalls

• Assuming all M.2 slots are equal. CPU-direct Gen5 x4 and chipset Gen4 x2 slots perform very differently.
• Ignoring the electrical width of x16 slots. A second x16 slot is often wired for x4.
• Missing the USB4/M.2 share. Enabling USB4 can halve an M.2 slot's bandwidth.
• Buying for maximum slots, not your device list. Unused slots cost nothing, but shared lanes you do use cost performance.

What to do next

Pick your two or three candidate boards, trace the lane map for each against your planned device list, and only then compare the remaining spec differences. If a lane conflict appears in your configuration, either change the board or change which slot you populate.

You need 16 PCIe lanes for a graphics card and 4 lanes for a primary NVMe SSD — 20 lanes minimum. Modern CPUs supply roughly 20–24 usable lanes, and the chipset adds more for extra drives and cards. The catch is that some slots share lanes, so populating one can slow another (a shared M.2 slot can drop your GPU slot from x16 to x8). Before buying, check the specific board's lane map rather than assuming every slot runs at full speed.

What PCIe lanes actually are

PCIe lanes are data pathways connecting your CPU and chipset to components like GPUs, NVMe SSDs, and expansion cards. Each lane is two pairs of wires — one pair sending, one receiving — so data moves in both directions at once.

Slots are labeled by lane count, and more lanes mean more bandwidth:

A physical x16 slot doesn't guarantee x16 bandwidth. A slot can be wired for x4 or x8 even though it's long enough to hold a full-size card — which is exactly why the lane map matters.

How many lanes you get, and from where

Two sources supply lanes:

• CPU-direct lanes: typically 20–24 usable lanes on modern platforms. These are the fastest and lowest-latency, and they usually feed the primary GPU slot and the first M.2 slot.
• Chipset lanes: the chipset expands connectivity for additional M.2 slots, PCIe slots, USB, and networking. These lanes share an uplink back to the CPU, so heavy simultaneous load across many chipset devices can contend for that link.

The common baseline is 16 lanes for the GPU plus 4 for the primary NVMe SSD. That's 20 lanes, which fits inside the CPU's direct allocation on most boards. Everything beyond that — a second or third NVMe drive, a capture card, a 10G network card — generally routes through chipset lanes.

How lane sharing changes your build

Lane sharing is when two or more slots use the same PCIe lanes. Only one can use them at full width at a time, so the board splits or switches them.

A concrete example from a mapped X870E board: the GPU slot runs Gen5 x16, and an M.2 slot shares lanes with USB4. Install a drive in that shared M.2 slot and the GPU slot can drop from x16 to x8. The board isn't broken — it's a design trade-off manufacturers make to fit more connectivity into a fixed lane budget.

This is why "how many M.2 slots does it have?" is the wrong question on its own. The right question is "how many M.2 slots can I populate without slowing my GPU or another slot I care about?"

When you need more than the baseline

You're pushed onto chipset lanes — and into potential sharing — when you add:

• A second, third, or fourth NVMe drive
• A capture card
• A 10G or faster network card
• A second GPU or compute card

If your build is one GPU plus one NVMe drive, lane sharing is rarely a problem. If you're planning three NVMe drives plus a capture card, you need to trace the routing before you buy.

How to check a specific board before buying

1. Find the board's lane map. Tools like MoboMaps publish interactive maps that trace every lane from its source (CPU or chipset) to each PCIe and M.2 slot, across 217 mapped boards.
2. Identify which slots are CPU-direct vs. chipset. CPU-direct slots are the ones you want for your GPU and primary SSD.
3. Toggle your planned devices. A bandwidth conflict checker shows which slots share lanes and flags conflicts — for example, "M.2_3 shares lanes with USB4."
4. Confirm the resulting widths. Check that your GPU still runs at the width you expect once your drives are installed.
5. Compare boards on the same dimensions. Stack two candidates side by side and compare M.2 slot count, PCIe x16 slots, USB4, and lane routing — not just price and power stages.

Catching a conflict at step 3 costs nothing. Catching it after you've built means returning a drive or accepting a slower GPU link.

Quick answers

How many PCIe lanes do I need? 20 minimum: 16 for the GPU, 4 for the primary NVMe SSD. Add chipset lanes for anything beyond that.

Will a second NVMe drive slow my GPU? Only if that M.2 slot shares lanes with the GPU slot. Many do; many don't. Check the map for your exact board.

Does a longer slot mean more bandwidth? No. A physical x16 slot can be wired for x4. The lane map tells you the real width.

Is lane sharing a defect? No — it's a trade-off. The board gives you more slots than it has lanes, and switches between them.

A PCIe lane map is a diagram of where every PCIe lane on a motherboard comes from and where it goes — which slots are wired directly to the CPU, which run through the chipset, how many lanes each slot gets, and which slots share lanes with each other. You read one by tracing each slot back to its source (CPU or chipset), noting the lane count and PCIe generation, then checking for shared lanes that drop bandwidth when two slots are populated at once. You need this before buying a board if your build combines a GPU, multiple NVMe drives, or expansion cards, because lane sharing can silently cut your GPU from x16 to x8 or split an M.2 slot's bandwidth.

The two lane sources: CPU-direct vs chipset

Every slot on a board traces back to one of two places.

CPU-direct lanes are the fastest and most predictable. As MoboMaps' PCIe Basics section puts it, CPUs provide roughly 20–24 usable PCIe lanes. The typical split is 16 lanes for the primary graphics slot plus 4 lanes for the first M.2 slot — 20 lanes minimum for a standard build. These lanes don't compete with chipset devices.

Chipset lanes expand connectivity beyond what the CPU offers. Extra M.2 slots, secondary PCIe slots, USB4 controllers, and network cards hang off the chipset. They share the chipset's uplink to the CPU, so heavy simultaneous use across chipset devices can contend for that link.

On a lane map, you'll see this distinction labeled directly. In the MoboMaps example for the ASUS ROG Crosshair X870E Hero (X870E, AM5, ATX), the map reads:

The pattern is common: the top x16 slot and first M.2 come straight from the CPU, everything else routes through the chipset.

Reading slot labels: lane count and generation

Slot labels tell you two things — how many lanes (x1, x4, x8, x16) and which PCIe generation (Gen3, Gen4, Gen5). Both matter for bandwidth.

Lane count sets the width of the pathway. A x16 slot has 16 lanes; a x4 slot has 4. More lanes means more bandwidth, which is why GPUs want x16 and NVMe drives typically use x4.

Generation sets the speed per lane. Each generation roughly doubles per-lane bandwidth over the previous one, so a Gen4 x4 slot moves about twice the data of a Gen3 x4 slot. A Gen5 x4 M.2 slot is faster still.

The practical takeaway: a Gen4 x4 M.2 slot and a Gen5 x4 M.2 slot have the same lane count but different ceilings. When you read a lane map, check both numbers — a slot labeled "Gen4 x4" is not the same as "Gen5 x4" even though both are x4.

Tracing lanes from source to slot

Reading a lane map means following each slot back to its origin and confirming the path makes sense for your parts.

1. Find your GPU slot. It's usually the top x16 slot, CPU-direct, Gen4 or Gen5. Confirm it's x16 and not shared in a way that drops it.
2. Find your primary NVMe slot. Typically CPU-direct x4, Gen4 or Gen5. This is your fastest storage path.
3. List every additional slot you plan to populate — second M.2, capture card, network card, second GPU.
4. Trace each one to its source. Chipset slots are fine for most add-in cards; just know they share the chipset uplink.
5. Check for shared lanes on any slot you're filling (next section).

MoboMaps describes this as tracing "every lane from source to slot" so you can "see exactly how PCIe lanes route from CPU and chipset to every PCIe and M.2 slot."

Spotting lane sharing and bandwidth conflicts

Lane sharing is when two or more slots use the same physical lanes. Populate one, and the other loses bandwidth — or the shared slot itself drops to fewer lanes.

MoboMaps gives two concrete examples of how this shows up:

• GPU drops from x16 to x8. Installing an NVMe drive in a shared M.2 slot can cut the graphics slot's lanes in half. The GPU still works, but with half the bandwidth.
• M.2 shares with USB4. In the Crosshair X870E Hero example, the performance checker flags "M.2_3 shares lanes with USB4" — a bandwidth conflict detected when both are in use.

The site frames this as a design trade-off: manufacturers route lanes to give you more slots, but not all slots can run at full speed simultaneously.

To catch conflicts, toggle the devices you plan to install and watch which slots change state. MoboMaps' checker shows slots as OK, Empty, or Shared, and surfaces a conflict count — "1 bandwidth conflict detected" in the example above. The goal, as the site puts it, is to "catch conflicts before you buy — not after you build."

A concrete conflict scenario

Say you want a Gen5 GPU in the top slot, two Gen4 NVMe drives, and a USB4 device. On a board where M.2_3 shares lanes with the USB4 controller, running both means one of them loses bandwidth. The lane map shows this before you order parts — you'd either move the second NVMe to a non-shared M.2 slot or accept the trade-off.

Using a lane map before you buy

A lane map answers one question: does my exact combination of parts fit without unexpected bandwidth loss?

Check these before committing to a board:

• GPU slot stays x16 with your storage and add-in cards installed.
• Primary M.2 is CPU-direct and runs at the generation you paid for.
• Every additional M.2 you'll populate is either non-shared or shares with something you won't use.
• Chipset uplink isn't oversubscribed by too many high-bandwidth chipset devices at once.

MoboMaps maps 217 boards and offers side-by-side comparison — stacking two boards to compare specs like M.2 slot count, PCIe x16 slots, USB4 ports, and power stages at a glance. In its example, the Crosshair X870E Hero (18 power stages, 5 M.2 slots, 2 USB4, $699) sits next to the ASRock X870E Taichi (24 power stages, 4 M.2 slots, 2 USB4, $499). Comparing lane maps side by side shows which board routes lanes the way your build needs.

The site describes itself as free, forever, with no pricing signals for paid tiers in the available material. It also lists a live YouTube Q&A for motherboard questions, which is a support channel rather than a product feature.

Quick reference

An Intel Z890 board gives you a CPU-attached x16 slot (Gen5 on most boards) plus a set of CPU and chipset M.2/PCIe slots whose lane counts and generations vary by model — so the only reliable way to know what you get is to read that specific board's lane map. Z890 is an LGA1851 platform paired with Core Ultra (Series 2) processors, and the CPU-to-chipset link is what feeds every chipset-attached slot. Before you buy or install a second NVMe drive, check whether the slot you want shares lanes with your GPU slot or a USB4 port.

What Z890 provides at the platform level

Two sources of lanes matter on any Z890 board:

• CPU lanes — a fixed pool that feeds the primary x16 graphics slot and typically the first M.2 slot. These are the fastest lanes (Gen5 on the graphics slot for most Z890 boards) and are not shared with the chipset.
• Chipset lanes — the Z890 chipset expands the CPU's uplink into additional PCIe and M.2 slots, plus USB, SATA, and networking. Chipset-attached slots are usually Gen4 or Gen3 and are the ones most likely to share lanes with each other.

The practical consequence: your GPU and primary SSD sit on dedicated CPU lanes and rarely conflict. Everything you add beyond that — second and third NVMe drives, capture cards, extra network cards — competes for chipset lanes and is where sharing shows up.

Typical Z890 lane layout

Most Z890 boards follow a recognizable pattern, though exact counts differ:

The exact mapping is board-specific. Two Z890 boards at similar prices can route their chipset lanes completely differently, which is why comparing spec sheets alone isn't enough.

How lane sharing works on Z890

Lane sharing means two or more slots draw from the same physical lanes. Populating one reduces or disables the other. On Z890 boards the common cases are:

• M.2 slot shares with a PCIe slot. Installing an NVMe drive in M.2_3 may disable a secondary PCIe x4 slot entirely, or drop it to x1.
• M.2 slot shares with USB4. A chipset M.2 slot may lose lanes when the board's USB4 controller is active, or vice versa.
• GPU slot drops to x8. Less common on Z890 than on some AMD boards, but it happens when a second CPU-attached slot is populated.

What drops is bandwidth, not compatibility — the device still works, just with fewer lanes. An SSD moved from x4 to x2 loses roughly half its sequential throughput, which is why a drive that benchmarks fine alone can feel slow once a second device is installed.

Reading a PCIe lane map for a specific Z890 board

A lane map traces every lane from its source (CPU or chipset) to its destination slot. To use one:

1. Find your board. Look it up by model name in a lane-map database or the board's manual block diagram.
2. Identify the CPU-attached slots first. These are your GPU slot and usually M.2_1 — they're independent and safe to populate.
3. Trace each chipset slot back to its source. Note which slots list a shared source.
4. Toggle the devices you plan to install. A conflict checker shows which slots degrade or disable when your intended combination is populated.
5. Confirm against the manual. Manufacturer manuals list the same sharing rules in the storage and expansion tables.

The goal is to catch a conflict before you buy the drive, not after you build and wonder why a slot went dark.

Checking for conflicts before you buy

Run this check against any Z890 board you're considering:

• List every device you'll install: GPU, each NVMe drive, any expansion card.
• For each M.2 slot you plan to use, confirm whether it shares lanes with a PCIe slot or USB4.
• Confirm the GPU slot stays at x16 with your intended M.2 population.
• Check whether populating a second or third M.2 disables a PCIe slot you need.
• If you rely on USB4, verify which M.2 slots remain usable with it active.

If your planned build is one GPU plus one or two NVMe drives, most Z890 boards handle it without conflict. Conflicts appear when you add a third or fourth drive, a capture card, or a 10GbE card alongside a full M.2 loadout.

What to compare between Z890 boards

When comparing Z890 motherboards, use the same dimensions across candidates:

• CPU-attached M.2 count — how many Gen5/Gen4 slots run directly off the CPU without sharing.
• Chipset M.2 count and generation — how many additional drives you can add and at what speed.
• Sharing rules — which slots disable or degrade when others are populated.
• USB4 presence and cost — USB4 ports often consume lanes that would otherwise feed an M.2 slot.
• Secondary PCIe slot behavior — whether it survives a full M.2 loadout.

A board with more M.2 slots isn't automatically better if half of them conflict with each other. A board with fewer, non-sharing slots can be the better choice for a build that needs predictable bandwidth.

FAQ

How many PCIe lanes does Z890 provide? The CPU provides a fixed pool (typically 20–24 usable lanes) for the graphics slot and primary M.2, and the Z890 chipset expands that into additional PCIe, M.2, USB, and SATA connectivity. Exact counts depend on the board.

Does installing an M.2 drive slow down my GPU on Z890? Usually not — the primary GPU slot and first M.2 slot run on separate CPU lanes. It can happen on boards where a second CPU-attached slot shares with the graphics slot, so check the lane map for your specific model.

How do I know if two slots share lanes? The board's manual and its PCIe lane map both list sharing rules. A conflict checker that lets you toggle installed devices shows the effect directly.

Is lane sharing a defect? No — it's a design trade-off. Manufacturers route a limited number of chipset lanes across more slots than there are lanes to go around, so some slots must share.

An AMD X870E board gives you a CPU-direct Gen5 x16 slot for the GPU plus a CPU-direct Gen5 x4 M.2 slot, with everything else routed through the chipset — and that "everything else" is where lane sharing shows up. Before buying, trace the specific board's lane map and toggle your planned devices, because two X870E boards with identical slot counts can behave very differently once you populate the second or third M.2 slot. For example, on the ASUS ROG Crosshair X870E Hero, M.2_3 shares lanes with USB4, so filling that slot costs you USB4 bandwidth rather than GPU lanes.

How X870E splits its lanes

X870E lane routing follows a consistent pattern across boards:

• CPU-direct: a Gen5 x16 slot (GPU) and a Gen5 x4 M.2 slot. These are the lanes that matter most for peak performance and are rarely shared.
• Chipset (X870E): the remaining PCIe and M.2 slots, plus USB4. The chipset fans these lanes out to multiple slots, which is exactly where sharing is introduced.

The Crosshair X870E Hero map shows this split clearly:

Note that M.2_2 is CPU-direct but Gen4 — CPU-direct doesn't automatically mean Gen5. Check the generation per slot, not just the source.

Why X870E boards share lanes

Lane sharing is a design trade-off, not a defect. The chipset has a fixed lane budget, so manufacturers route the same lanes to two or more slots and let you use one or the other. The most common X870E conflicts:

• M.2 slot vs. USB4 — populating a shared M.2 slot reduces or disables USB4 bandwidth (this is the Crosshair X870E Hero's M.2_3 behavior).
• Secondary PCIe slot vs. M.2 slots — installing a card in PCIEX16_2 can drop an M.2 slot to fewer lanes or disable it.
• GPU slot dropping to x8 — less common on X870E than on some Intel platforms, but still possible on boards that split the CPU x16.

The practical consequence: a board advertising five M.2 slots may only run all five at full width if you sacrifice USB4 or a PCIe slot. The spec sheet rarely tells you this; the lane map does.

How to trace a specific X870E board's lane map

1. Open the board's lane map. On MoboMaps, search the board by name (e.g., "ASUS ROG Crosshair X870E Hero") or browse the X870E category. The map traces every lane from its source (CPU or chipset) to its destination slot.
2. Read each slot's source, generation, and width. Confirm which slots are CPU-direct and which come from the chipset. CPU-direct slots are your performance-critical ones.
3. Toggle your planned devices. Use the performance checker to mark which M.2 slots and PCIe slots you intend to populate. The tool flags bandwidth conflicts — for example, "1 bandwidth conflict detected: M.2_3 shares lanes with USB4."
4. Check the resulting state of each slot. The checker shows whether each slot is OK, Shared, or degraded. A slot marked "Shared" means populating it affects another slot you may also want.

The expected result: you see exactly which of your planned components compete for the same lanes before you buy, rather than discovering it after the build.

What to compare between X870E boards

When choosing between two X870E boards, compare on these dimensions — the same ones the board comparison view uses:

For example, comparing the ROG Crosshair X870E Hero ($699) against the ASRock X870E Taichi ($499) side by side shows the Hero with 5 M.2 slots and the Taichi with 4, both with 2 USB4 ports and 2 PCIe x16 slots. The lane maps then tell you which slots on each board share — the spec counts alone don't.

The short version

X870E gives you a clean CPU-direct Gen5 x16 plus Gen5 x4 M.2, and a chipset that fans out the rest. Lane sharing lives in the chipset-fed slots, most often between an M.2 slot and USB4 or a secondary PCIe slot. Trace the specific board's lane map, toggle your planned devices, and resolve conflicts before purchase — the board with the most slots isn't necessarily the one that runs your configuration without compromise.