Compare FPGA devices

No architecture-neutral logic-cell count is established. The vendor's nominal LUT4 quantity is retained with its precision and source in lut4_nominal, not converted into AMD logic cells.

No architecture-neutral logic-cell count is established. The vendor's nominal LUT4 quantity is retained with its precision and source in lut4_nominal, not converted into AMD logic cells.

ECP5 uses LUT4 primitives. Wider logic functions can combine LUT4s, but that is not a source-defined native LUT6 count.

ECP5 uses LUT4 primitives. Wider logic functions can combine LUT4s, but that is not a source-defined native LUT6 count.

Source reports embedded memory in kb (kilobits), excluding the separate Distributed RAM Bits row; each EBR is 18 kb.

Source reports embedded memory in kb (kilobits), excluding the separate Distributed RAM Bits row; each EBR is 18 kb.

The generic field means 36 Kb blocks. ECP5 sysMEM EBR blocks are 18 Kb; sysmem_18kb_blocks retains their physical granularity. No artificial 36 Kb block count is supplied.

The generic field means 36 Kb blocks. ECP5 sysMEM EBR blocks are 18 Kb; sysmem_18kb_blocks retains their physical granularity. No artificial 36 Kb block count is supplied.

The generic application field denotes DSP slices. The directly printed native count is 18 x 18 multipliers, retained in embedded_18x18_multipliers; separately derived sysDSP slice/block counts retain their own native definitions.

The generic application field denotes DSP slices. The directly printed native count is 18 x 18 multipliers, retained in embedded_18x18_multipliers; separately derived sysDSP slice/block counts retain their own native definitions.

Manufacturer reports 44 k LUTs in Table 1.1, normalized as k x 1000 for the numeric field. Native primitive is LUT4 (p. 16). This is a nominal vendor figure, not a verified exact physical LUT count or an AMD logic-cell equivalent; do not use density code 45F as the count.

Manufacturer reports 44 k LUTs in Table 1.1, normalized as k x 1000 for the numeric field. Native primitive is LUT4 (p. 16). This is a nominal vendor figure, not a verified exact physical LUT count or an AMD logic-cell equivalent; do not use density code 45F as the count.

Directly printed multiplier count. One sysDSP slice supports two 18 x 18 multipliers (or four 9 x 9); one physical sysDSP block contains two slices (TN-02205-1.3 p. 6). This number is not a DSP-slice/block count and is not equivalent to AMD DSP48E1.

Directly printed multiplier count. One sysDSP slice supports two 18 x 18 multipliers (or four 9 x 9); one physical sysDSP block contains two slices (TN-02205-1.3 p. 6). This number is not a DSP-slice/block count and is not equivalent to AMD DSP48E1.

Native block granularity; keep separate from app's 36 Kb block field.

Native block granularity; keep separate from app's 36 Kb block field.

Derived as multiplier count / 2. A slice has two 18 x 18 multipliers; this is distinct from logic/PFU slices.

Derived as multiplier count / 2. A slice has two 18 x 18 multipliers; this is distinct from logic/PFU slices.

Derived as DS-02012 multiplier count / 4: two 18-bit multipliers per slice, two slices per block. A DSP block is not a DSP slice.

Derived as DS-02012 multiplier count / 4: two 18-bit multipliers per slice, two slices per block. A DSP block is not a DSP slice.

Separate from EBR capacity; do not add to bram without changing that field's definition.

Separate from EBR capacity; do not add to bram without changing that field's definition.

Package-specific single-ended user I/O total, excluding dedicated SerDes channels. General-purpose I/O totals do not equal total package terminals.

Package-specific single-ended user I/O total, excluding dedicated SerDes channels. General-purpose I/O totals do not equal total package terminals.

The package-specific SerDes Channels/I/O Count cell gives exposed channels. Table 2.14 counts duals (two channels per dual); the pin-summary SerDes row counts pins, not channels.

The package-specific SerDes Channels/I/O Count cell gives exposed channels. Table 2.14 counts duals (two channels per dual); the pin-summary SerDes row counts pins, not channels.

BG554 is explicitly defined as 554-ball lead-free caBGA. The full MPN is independently established by the literal ordering row.

BG554 is explicitly defined as 554-ball lead-free caBGA. The full MPN is independently established by the literal ordering row.

Actual ball count explicitly printed in the pin-summary table and selected literal ordering row, independently checked against the package-drawing title; not inferred from suffix digits.

Actual ball count explicitly printed in the pin-summary table and selected literal ordering row, independently checked against the package-drawing title; not inferred from suffix digits.

Package body width x length, basic dimension (BSC), not PCB footprint/keep-out or height. Cross-checked against DS-02012 Table 1.1.

Package body width x length, basic dimension (BSC), not PCB footprint/keep-out or height. Cross-checked against DS-02012 Table 1.1.

Nominal ball-to-ball pitch (BSC). Depopulated ball arrays still have this underlying pitch.

Nominal ball-to-ball pitch (BSC). Depopulated ball arrays still have this underlying pitch.

Literal ordering-row grade. Grades describe timing bins within this family, not a single application fMAX or a cross-vendor speed ranking. ECP5-5G ordering rows in this revision list only -8.

Literal ordering-row grade. Grades describe timing bins within this family, not a single application fMAX or a cross-vendor speed ranking. ECP5-5G ordering rows in this revision list only -8.

Recommended operating junction temperature Tj. This is not ambient Ta, storage temperature, or the absolute maximum junction rating.

Recommended operating junction temperature Tj. This is not ambient Ta, storage temperature, or the absolute maximum junction rating.

The grade is read from this exact literal full MPN's ordering row, then joined to its specified junction operating range.

The grade is read from this exact literal full MPN's ordering row, then joined to its specified junction operating range.

Lattice names the core rail VCC; stored under app vccint only as a generic core-voltage field. This is the recommended core range, not SerDes VCCA, VCCIO, an absolute maximum, or a full board power-supply specification.

Lattice names the core rail VCC; stored under app vccint only as a generic core-voltage field. This is the recommended core range, not SerDes VCCA, VCCIO, an absolute maximum, or a full board power-supply specification.

Application fMAX depends on implementation, path, loading, speed grade and operating conditions. No single application clock is established by the speed grade or PLL limit.

Application fMAX depends on implementation, path, loading, speed grade and operating conditions. No single application clock is established by the speed grade or PLL limit.

Application power depends on design activity, resources, I/O, voltage and temperature; not determined from marketing/typical static-current figures.

Application power depends on design activity, resources, I/O, voltage and temperature; not determined from marketing/typical static-current figures.

Recommended operating junction-temperature minimum.

Recommended operating junction-temperature minimum.

Recommended operating junction-temperature maximum.

Recommended operating junction-temperature maximum.

Tj is explicitly stated by the manufacturer; no ambient range is established.

Tj is explicitly stated by the manufacturer; no ambient range is established.

No current official MPN lifecycle status was established. An ordering-table entry or live family webpage is not an Active-status statement.

No current official MPN lifecycle status was established. An ordering-table entry or live family webpage is not an Active-status statement.

No current stock/inventory evidence was obtained. The official ordering-table entry does not establish availability.

No current stock/inventory evidence was obtained. The official ordering-table entry does not establish availability.